Soup search results

For discussion of specific patterns or specific families of patterns, both newly-discovered and well-known.
User avatar
wwei23
Posts: 1932
Joined: May 22nd, 2017, 6:14 pm
Location: The (Life?) Universe

Re: Soup search results

Post by wwei23 » October 16th, 2020, 11:25 pm

MathAndCode wrote:
October 16th, 2020, 11:21 pm
The first HWSS saved the second HWSS. Otherwise, this would have happened.
You're right! That's actually pretty cool, not to mention how fast it forms.
@Hunting so we're making a referential signature chain?

MathAndCode
Posts: 1431
Joined: August 31st, 2020, 5:58 pm

Re: Soup search results

Post by MathAndCode » October 16th, 2020, 11:28 pm

wwei23 wrote:
October 16th, 2020, 11:25 pm
You're right! That's actually pretty cool, not to mention how fast it forms.
Yes; I supposed that the first HWSS was a team player. However, the fast formation makes the soup useless for most attempts at an improved glider syntehsis.
Marine has a non-centered Moore neighborhood and at least one elementary gun with a very high period (11,843).
I'm in the process of reducing the glider cost of universal construction.

MikeP
Posts: 102
Joined: February 7th, 2010, 9:51 am
Location: Ely, Cambridgeshire, UK

Re: Soup search results

Post by MikeP » October 23rd, 2020, 7:47 am

After just over 100bn soups in T29x33 I have the following period statistics:

1: 23339265523
2: 77087261915
3: 20124173
4: 25796
5: 202
6: 31545234
8: 28743
10: 462
12: 10
14: 404
15: 216520
24: 8
30: 460690
46: 1
116: 6581153
132: 4711752
348: 1121
488: 4
660: 7
1194: 119
1740: 23
1914: 409
3828: 33775731

Total: 100524000000

The only new pattern I haven't posted yet is this p488 wick:

Code: Select all

#CXRLE Pos=-20,-61 Gen=545
x = 23, y = 115, rule = B3/S23:T50,132
9b2o$9bobo$9b3o6bo$7bobo7b2o$b2o5bo8bo$o4b2ob2o11b2o$bob3o10b3ob2o$b5o
bob2o9bo$6bobo$8b2o$8b3ob4o$8b2o3bobo$9bo3b2o$9b2ob3o$9bo2b3o$8b3o18$
9b2o$9bobo$9b3o6bo$7bobo7b2o$b2o5bo8bo$o4b2ob2o11b2o$bob3o10b3ob2o$b5o
bob2o9bo$6bobo$8b2o$8b3ob4o$8b2o3bobo$9bo3b2o$9b2ob3o$9bo2b3o$8b3o18$
9b2o$9bobo$9b3o6bo$7bobo7b2o$b2o5bo8bo$o4b2ob2o11b2o$bob3o10b3ob2o$b5o
bob2o9bo$6bobo$8b2o$8b3ob4o$8b2o3bobo$9bo3b2o$9b2ob3o$9bo2b3o$8b3o18$
9b2o$9bobo$9b3o6bo$7bobo7b2o$b2o5bo8bo$o4b2ob2o11b2o$bob3o10b3ob2o$b5o
bob2o9bo$6bobo$8b2o$8b3ob4o$8b2o3bobo$9bo3b2o$9b2ob3o$9bo2b3o$8b3o!
From this soup (and three others):

Code: Select all

#CXRLE Pos=-14,-16
x = 29, y = 33, rule = B3/S23:T29,33
2b3o2b4o3b2o3bo2bob2o2bo$6o2bo3b2o5b2o2bobo2bo$obo2bo3b5o2b2obob2o3bo$
2b6o3b2o7bobo4bo$3obobob3o5b2o6b5o$obob3o2b6o5b4o4bo$2o3bo2bob2ob2o2bo
b2ob3obo$bob2o9b2o3bo2b2obobo$2obob2obo6bo2b3o2bob2obo$3o2bob4obo2bobo
2bo2bobo$6b4obob2obo2bobob4o$2obo3bobobob5ob2o2bo3b2o$2ob5o2bo4b2o2bo
2b2o3bo$o3bob4ob2obob2o2bo4bo$3obobo2b2o6b2o3b2o3bo$ob5o6b2o2bo4b3obob
o$obob3obobobo2b2o4b4obo$o3bob3ob3o3bo4b3obobo$2b4ob3o3b2ob7ob4o$6ob2o
bo4bobo4b2ob2obo$2bo2b2ob2o3bo4b2obob2ob2o$obob2o3b2o3b2ob3o2bo2b2o$bo
b5obo3b3obo3bobo3bo$2bo3bo5bo2b2obobobob3o$o3bob2obo2b2o2bo4b2o2b4o$5o
2b2o2bo2bo2b5obob4o$o2b3o2bobo2bob2ob2o3b3obo$o5b2o2b7ob4o2bo2b2o$obob
6obo2bo6bo2b2obo$b4o2b3obo2bob3obobob2o2bo$2b4o2b2obobo3bo2b2o3bob2o$o
6b3o3bob2o2b3o2bo3bo$ob3obo2bobob10obo2bo!

User avatar
BlinkerSpawn
Posts: 1977
Joined: November 8th, 2014, 8:48 pm
Location: Getting a snacker from R-Bee's

Re: Soup search results

Post by BlinkerSpawn » October 24th, 2020, 9:10 pm

MikeP wrote:
October 23rd, 2020, 7:47 am
The only new pattern I haven't posted yet is this p488 wick:

Code: Select all

#CXRLE Pos=-20,-61 Gen=545
x = 23, y = 115, rule = B3/S23:T50,132
9b2o$9bobo$9b3o6bo$7bobo7b2o$b2o5bo8bo$o4b2ob2o11b2o$bob3o10b3ob2o$b5o
bob2o9bo$6bobo$8b2o$8b3ob4o$8b2o3bobo$9bo3b2o$9b2ob3o$9bo2b3o$8b3o18$
9b2o$9bobo$9b3o6bo$7bobo7b2o$b2o5bo8bo$o4b2ob2o11b2o$bob3o10b3ob2o$b5o
bob2o9bo$6bobo$8b2o$8b3ob4o$8b2o3bobo$9bo3b2o$9b2ob3o$9bo2b3o$8b3o18$
9b2o$9bobo$9b3o6bo$7bobo7b2o$b2o5bo8bo$o4b2ob2o11b2o$bob3o10b3ob2o$b5o
bob2o9bo$6bobo$8b2o$8b3ob4o$8b2o3bobo$9bo3b2o$9b2ob3o$9bo2b3o$8b3o18$
9b2o$9bobo$9b3o6bo$7bobo7b2o$b2o5bo8bo$o4b2ob2o11b2o$bob3o10b3ob2o$b5o
bob2o9bo$6bobo$8b2o$8b3ob4o$8b2o3bobo$9bo3b2o$9b2ob3o$9bo2b3o$8b3o!
That one's interesting: 102 generations after that snapshot it settles into a dove, a blinker, and what soon turns into a block.
LifeWiki: Like Wikipedia but with more spaceships. [citation needed]

Image

User avatar
Extrementhusiast
Posts: 1858
Joined: June 16th, 2009, 11:24 pm
Location: USA

Re: Soup search results

Post by Extrementhusiast » October 28th, 2020, 1:06 am

mniemiec wrote:
September 17th, 2020, 4:29 pm
Entity Valkyrie 2 wrote:
September 13th, 2020, 6:21 am
I think this P4 needs a name, since it is an edgy domino sparker, and it is statorless. ..
dvgrn wrote:
September 13th, 2020, 6:34 am
Mark Niemiec's database has it as "24-bit P4 #1", no synthesis; pentadecathlon.com calls it 24P4.1.
Catagolue has one C2_4 soup for xp4_g880anziaw5kze5011 but the oscillator shows up early, so converting to a glider recipe looks tricky.
I have generally been assigning canonical names (like "15.60") for still-lifes that have been exhaustively searched, and ad-hoc names (like "45-bit still-life #1") for ones that have not (or, more recently, for 25+-bit ones that have been seached, but for which I don't have the list myself). For oscillators, I often used similar ad-hoc names like "24-bit period four oscillator #1", but I have been slowly trying to convert these and spaceships to the Pentadecathlon nomenclature (like 24P4.1 or 20P7H1V0.1), or its equivalent when the patterns aren't actually on the Pentadecathlon list. Unfortunately, since the suffix assignments are arbitrary, there's no objective way to name them, but this is rarely a problem as the number of oscillators and spaceships is usually fairly small.
Entity Valkyrie 2 wrote:
September 17th, 2020, 6:55 am
Is it possible to synthesize these inserts between the blocks? ...
(By the way, you will likely have an easier time if you replace the two separate one-bit sparks at the bottom with a domino-spark; the results are the same in the next generation).

It might be possible, but I doubt it would be easy. The 4-bit spark at the bottom left is basically an "insert a block via broken toad" method; I know of several of these; unfortunately, about half of them use a predecessor that would hit the block first, and all of them require gliders that come from the direction of the block. I suspect that a large random 3- or 4-glider search might find some suitable ways to make this.
Generation 5 was all I needed (along with a few other things not part of that soup):

Code: Select all

x = 492, y = 40, rule = B3/S23
188bobo185bo19bo$122bo66b2o186bo17bo$121bo67bo88bo92bo3b3o17b3o$121b3o
155b2o17bo73bo$195bo82b2o17bo72b3o$108bo87b2o99b3o$106b2o87b2o2bo89bo$
104bo2b2o90bobo87bobo$102bobo94b2o88b2o$103b2o9bo375bo$113bo375bo$113b
3o175bo4bo192b3o$97b3o192bob2o190b2o$290b3o2b2o89bo6bo4b2o86bobo$11bob
o5bo75bo5bo283bobo4bo5bobo86bo$11b2o4b2o76bo5bo95b2o84b2o98bo2bo5b3o3b
o$12bo5b2o75bo5bo95bo2b2o81bobob2o93bob2o89bob2o$14bo183bobo84bobo94bo
bo9bo75bo3bo$14b2o81b3o3b2o92b2obo83b2obo93b2obo8b2o74b2o7bo$13bobo73b
2o11bo2bo90bo4b2o80bo4b2o90bo4b2o6bobo74bo2b2ob2o$4bobo81bo2bo11b2o92b
2o4bo80b2o4bo81bobo6b2o4bo78b2ob2o2bo$4b2o83b2o3b3o102bob2o83bob2o83b
2o8bob2o78bo7b2o$5bo193bobo84bobo84bo9bobo83bo3bo$2o5bo84bo5bo99b2o2bo
82b2obobo91b2obo79b2obo$b2o4b2o83bo5bo102b2o86b2o78bo3b3o5bo2bo$o5bobo
83bo5bo268bobo5bo4bobo73bo$277b2o2b3o84b2o4bo6bo73bobo$94b3o181b2obo
174b2o$78b3o196bo4bo169b3o$80bo373bo$79bo9b2o362bo$89bobo107b2o82b2o$
85b2o2bo108bobo81bobo$86b2o112bo2b2o79bo$85bo116b2o70b3o$204bo71bo17b
2o99b3o$70b3o202bo17b2o100bo$72bo137bo84bo74b3o17b3o3bo$71bo137b2o161b
o17bo$209bobo159bo19bo!
I Like My Heisenburps! (and others)

John Goodman
Posts: 31
Joined: December 13th, 2019, 10:00 am

Re: Soup search results

Post by John Goodman » November 14th, 2020, 8:42 pm

The second 49k methuselah has been discovered, this time in C1, by inspirehep@gmail.com. It lasts 49132 generations.

Code: Select all

x = 16, y = 16, rule = B3/S23
bbbbooobbbbbboob$
bobbbbbobbboobbo$
bobbobooobbbobbo$
oobboooobooboobo$
bboobbobboobobob$
ooooobboobobobbo$
bbbbobbbbobbbbbo$
boobbbboboooobbb$
booboooobooobobb$
boooobooooobbobb$
bbboobbbboobbobo$
obbobbbobooobbob$
oobboooboobboobb$
obobooobbobbbbbb$
obbbbbobbbbbbbob$
bbobbboboooboobb!

MathAndCode
Posts: 1431
Joined: August 31st, 2020, 5:58 pm

Re: Soup search results

Post by MathAndCode » November 14th, 2020, 9:09 pm

John Goodman wrote:
November 14th, 2020, 8:42 pm
The second 49k methuselah has been discovered, this time in C1, by inspirehep@gmail.com. It lasts 49132 generations.
Here is a version with fewer cells. (This is the drill, right?)

Code: Select all

x = 23, y = 28, rule = B3/S23
17b3o$19bo2$5bo$4bobo$4b2o$8b3o$7bo12bo$10bo8b2o$10bo7b2o$9b2o7bo$2o$bo$9bo$8b2o$8bo2$17b2o$17bobo$18bo3$12bo$11bobo7bo$12bo9bo$22bo$22bo$22bo!


Edit: I looked further, and there is a seven-cell dove. (I previously did not know of any dove sequence predecessors with fewer than eight cells.)



Another edit: I found a six-cell parent.

Code: Select all

x = 4, y = 5, rule = B3/S23
o$b2o$3bo$3bo$3bo!


Yet another edit: The 49k-generation methuselah forms a block on loaf and a beehive on beehive, but each is later destroyed.
Marine has a non-centered Moore neighborhood and at least one elementary gun with a very high period (11,843).
I'm in the process of reducing the glider cost of universal construction.

User avatar
dvgrn
Moderator
Posts: 7387
Joined: May 17th, 2009, 11:00 pm
Location: Madison, WI
Contact:

Re: Soup search results

Post by dvgrn » November 15th, 2020, 8:58 am

MathAndCode wrote:
November 14th, 2020, 9:09 pm
John Goodman wrote:
November 14th, 2020, 8:42 pm
The second 49k methuselah has been discovered, this time in C1, by inspirehep@gmail.com. It lasts 49132 generations.
Here is a version with fewer cells. (This is the drill, right?)
It's one of the drills, anyway. There's still no clear hierarchy for what makes a methuselah "better" or "worse", except we mostly agree that "natural" methuselahs (made with a good hash function like SHA256, so we don't know how to reverse engineer a hash value for a given pattern) are more noteworthy than artificially constructed variants. And smaller bounding boxes are better, and longer lifespans are better, and lower populations are better.

Sadly, if the last fifty years of discussion about these points is any guide, no two Lifenthusiasts will ever agree about exactly how to score these various measurements.

For example, your methuselah variant is only 44 cells instead of 113, but it lasts only 49038 ticks instead of 49132, and it's 23x28 instead of 16x16. Is that better or worse than this variant I just made with JLS, which has 113 cells like the original one, but it's only 19x19 and lasts one tick longer than the original, 49133 ticks?

Code: Select all

x = 19, y = 19, rule = B3/S23
11b2o$4bob2o6bobo$6b2o2bob2o2bo$5bo2bo4bo2bobo$bob3o3bo2b2o2bobo$obo3b
2o2bo2bob2o$2bo3b2o4bo2bobo$2bo2bo2bo6bo$b3ob3o3b2o$5b2ob2ob2obo$3bo3b
2o5bobo$2b2o4b2o2b2o2bo$2b2o2bo2b2o4bo$2bobobo3bo2b2o$bo5bo3b3o$b2o3bo
4b2o2bo$2b3o3bob2ob3o$bo6bo5bo$3bobo4b2obo2bo!
It's easy to make a metric that combines population, bounding box, stabilization time, etc., to make a single number that can be compared against other methuselahs' numbers. People have invented dozens of such metrics, so now we have way too many of them that then have to be calculated and maintained in the List_of_long-lived_methuselahs -- argh.

... After about five minutes thinking about it, I always just decide I don't really care, and stop before inventing my own improved metric and making the problem even worse.

Question for the Experts
By the way -- I can't seem to get the population constraints to work in JLS. I might have asked this before, but does anyone know if that's just an un-implemented feature in JLS, or if I'm doing something wrong somehow with Search > Options > Constraints > No more than N cells in each layer? Checking or unchecking the "excluding manually set cells" doesn't make any difference.

(Obviously I should really be using something like LLS instead, but given that I stop caring about the result within five minutes anyway, JLS is plenty good enough.)

User avatar
wwei23
Posts: 1932
Joined: May 22nd, 2017, 6:14 pm
Location: The (Life?) Universe

Re: Soup search results

Post by wwei23 » November 15th, 2020, 11:28 am

dvgrn wrote:
November 15th, 2020, 8:58 am
By the way -- I can't seem to get the population constraints to work in JLS. I might have asked this before, but does anyone know if that's just an un-implemented feature in JLS, or if I'm doing something wrong somehow with Search > Options > Constraints > No more than N cells in each layer? Checking or unchecking the "excluding manually set cells" doesn't make any difference.
Check the box on the left, the "No more than x on cells" one, not the "excluding manually set cells" one.
@Hunting so we're making a referential signature chain?

wildmyron
Posts: 1489
Joined: August 9th, 2013, 12:45 am
Location: Western Australia

Re: Soup search results

Post by wildmyron » November 17th, 2020, 12:55 pm

dvgrn wrote:
November 15th, 2020, 8:58 am
Question for the Experts
By the way -- I can't seem to get the population constraints to work in JLS. I might have asked this before, but does anyone know if that's just an un-implemented feature in JLS, or if I'm doing something wrong somehow with Search > Options > Constraints > No more than N cells in each layer? Checking or unchecking the "excluding manually set cells" doesn't make any difference.

(Obviously I should really be using something like LLS instead, but given that I stop caring about the result within five minutes anyway, JLS is plenty good enough.)
To impose a population constraint in JLS you need to choose the option Search > Options > Constraints > No more than N ON cells in generation 0, and as wwei23 mentioned, don't check "excluding manually set cells" (unless you account for them in the total). It's not possible to apply a total population constraint to the other generations in JLS. The other constraints applied to "Layers" are designed to constrain how many ON cells there are per row or column. They can be applied to any sorting order though, but it doesn't make much sense to apply a single population constraint to expanding circular rings (for example). When you use one of these options you can see the individual layers highlighted by tinting applied to the cell borders of each cell in a layer. They layers are alternately tinted black, blue and green (difficult to see on some monitors though).
The latest version of the 5S Project contains over 226,000 spaceships. There is also a GitHub mirror of the collection. Tabulated pages up to period 160 (out of date) are available on the LifeWiki.

User avatar
wwei23
Posts: 1932
Joined: May 22nd, 2017, 6:14 pm
Location: The (Life?) Universe

Re: Soup search results

Post by wwei23 » November 19th, 2020, 9:32 am

A 52-bit still life showed up in ikpx2_stdin, the only symmetry with it so far:

Code: Select all

x = 0, y = 0, rule = B3/S23
3bo$2b3ob3o$bobo5bo$3obo5bo$bob2o4b4o$6b2o5b2o$bo3bob3o3b3o$bo3b4o
4b2o$bo4b3ob2obo$2bobobo2bobo$3b2o3bobo$4bo3b2o$4bo$5b4o$5b3o$6bo!
It was found by calcyman.
@Hunting so we're making a referential signature chain?

Dylan Chen
Posts: 21
Joined: March 27th, 2020, 8:07 am
Contact:

Re: Soup search results

Post by Dylan Chen » November 24th, 2020, 10:57 pm

First C1 methuselah_49k soup

Code: Select all

x = 16, y = 16, rule = B3/S23
bbbbooobbbbbboob$
bobbbbbobbboobbo$
bobbobooobbbobbo$
oobboooobooboobo$
bboobbobboobobob$
ooooobboobobobbo$
bbbbobbbbobbbbbo$
boobbbboboooobbb$
booboooobooobobb$
boooobooooobbobb$
bbboobbbboobbobo$
obbobbbobooobbob$
oobboooboobboobb$
obobooobbobbbbbb$
obbbbbobbbbbbbob$
bbobbboboooboobb!
lasts 49132 gens, slightly shorter than G1 soup #1 which lasts 49768 gens

found by inspirehep@gmail.com(yeah, that's me) on 14 November 2020

MathAndCode
Posts: 1431
Joined: August 31st, 2020, 5:58 pm

Re: Soup search results

Post by MathAndCode » November 24th, 2020, 11:32 pm

Dylan Chen wrote:
November 24th, 2020, 10:57 pm
First C1 methuselah_49k soup
This has already been mentioned here, but congratulations again for finding it.
Marine has a non-centered Moore neighborhood and at least one elementary gun with a very high period (11,843).
I'm in the process of reducing the glider cost of universal construction.

Dylan Chen
Posts: 21
Joined: March 27th, 2020, 8:07 am
Contact:

Re: Soup search results

Post by Dylan Chen » November 25th, 2020, 4:13 am

MathAndCode wrote:
November 14th, 2020, 9:09 pm

Here is a version with fewer cells. (This is the drill, right?)

Code: Select all

x = 23, y = 28, rule = B3/S23
17b3o$19bo2$5bo$4bobo$4b2o$8b3o$7bo12bo$10bo8b2o$10bo7b2o$9b2o7bo$2o$bo$9bo$8b2o$8bo2$17b2o$17bobo$18bo3$12bo$11bobo7bo$12bo9bo$22bo$22bo$22bo!
This one is amazing. More natural than the "natural" 16x16 solid soup.

MathAndCode
Posts: 1431
Joined: August 31st, 2020, 5:58 pm

Re: Soup search results

Post by MathAndCode » November 25th, 2020, 7:06 pm

Dylan Chen wrote:
November 25th, 2020, 4:13 am
This one is amazing. More natural than the "natural" 16x16 solid soup.
It's the standard practice..
Marine has a non-centered Moore neighborhood and at least one elementary gun with a very high period (11,843).
I'm in the process of reducing the glider cost of universal construction.

Post Reply