Thread For Your Useless Discoveries

[Scorbie]

Useless conduits:

Code: Select all

x = 30, y = 15, rule = LifeHistory
7.2B$7.3B$6.4B$6.6B16.A$5.D6B15.A.A$4.2D6B14.2BAB$5.2D4B16.2B$5.7B13.
3B$3.9B9.A2.4B$.B.9B8.A.A.5B$2A5BC5B8.A.3BC2B$2A4B3C4B9.3B3C2B$.4B2C
4B11.2B2C2BDB$.3B.5B14.2B3D$8.B!

[gmc_nxtman]

Another nice blinkic seed: Generates a hat during its evolution, and later creates (according to my "knowledge" of still life naming conventions) a dock weld mango on block.

Code: Select all

x = 84, y = 76, rule = S23/B3
21b3o23$18bobob3o$19b2o$19bo22$3o2$66bo$66bo$66bo11$81b3o13$38b3o!

EDIT: Another one: Integral with boat.

Code: Select all

x = 75, y = 43, rule = B3/S23
2bo$obo71bo$b2o71bo$74bo$3b3o38$55b3o!

EDIT2: Yet another double jewel! Generates a Cis-hive on long hook, and later a snake, both of which survive.

Code: Select all

x = 73, y = 40, rule = B3/S23
35b2o$35b2o33$71b2o$2bo68b2o$obo$b2o2$3b3o!

3-blinker seed for Bi-loaf 2:

Code: Select all

x = 79, y = 42, rule = B3/S23
76b3o9$45b3o28$2bo$obo$b2o2$3b3o!

This reaction, and a semistable way of suppressing it:

Code: Select all

x = 42, y = 6, rule = Life
16b2o$bo14b2o14bo$obo5bo22bobo5bo$b2o5b3o21b2o5b3o$8bobo28bobo$10bo30b
o!

Catalytic pre-honey farm:

Code: Select all

x = 11, y = 5, rule = Life
2o$obo5bo$b2o5b3o$8bobo$10bo!

These two very similar diehards with very different histories:

Code: Select all

x = 36, y = 4, rule = Life
bo6bo18bo6b2o$b2o4bobo17b2o4bobo$2o5bobo16b2o5bobo$7b2o24b2o!

72 is a nice number....

Code: Select all

x = 72, y = 1, rule = Life
72o!

A one-time R-pentomino turner:

Code: Select all

x = 4, y = 7, rule = Life
2bo$3o$bo3$2b2o$2b2o!

Diehards:

Code: Select all

x = 46, y = 8, rule = B3/S23
6bo$2o4bo22bo13bo$bo3bobo19b2o14b3o$29bobo11bobo$31bo13bo2$29b2o13b2o$
30bo14bo!

[praosylen]

A completely useless catalyst:

Code: Select all

x = 15, y = 13, rule = B3/S23
5b2ob2o$2b2o2bobo2b2o$o2bobo3bobo2bo$2obob2ob2obob2o$3bo2bobo2bo$3b2o
5b2o$7bo$7bo$3bo2bo2bo$4bo7b2o$bo2bo2b3ob3o$bob9o$4bo2bo2bo2bo!

Edit: Slightly less useless catalyst (which is not saying much):

Code: Select all

x = 8, y = 11, rule = B3/S23
3bo$2bobo$3bo2$b5o$o2bo2bo$2o3b2o2$4bo$4b2o$3b2ob2o!

[gmc_nxtman]

Introducing... Eaters! As you've never seen them before.

Code: Select all

x = 39, y = 39, rule = B3/S23
2bo$obo16bo$b2o17b2o$19b2o2$23bo$24b2o$23b2o2bob2o$9b2o16b2obo2b2o$5b
2o2bobo20b2o$6b2o3bo22bo$5bo5b2o$37b2o$b2o10bo22b2o$2b2o8b2o24bo$bo10b
obo2$17bo$16b2o$16bobo5$o$b2o$2o14bo$16bobo$4bo11b2o$5b2o6b2o$4b2o6bob
o$12bo$10bobo6b2o$10b2o6b2o$7b2o11bo$6bobo$8bo14b2o$22b2o$24bo!

This is probably already known, I just wanted to get it out there for other people to see.

[dvgrn]

Introducing... Eaters! As you've never seen them before.

I'm surprised you didn't also include eater2s (eating gliders on four adjacent incoming lanes) and tub-with-tail eaters (eating glider from two different directions).

See Golly's Patterns/Life/Still-Lifes/eaters.rle, and the other two "eater" patterns in the same folder.

[gmc_nxtman]

Introducing... Eaters! As you've never seen them before.

I'm surprised you didn't also include eater2s (eating gliders on four adjacent incoming lanes) and tub-with-tail eaters (eating glider from two different directions).

See Golly's Patterns/Life/Still-Lifes/eaters.rle, and the other two "eater" patterns in the same folder.

Well, I just thought that the eater2 would be already known, but the other's wouldn't be. Also, what is the smallest way to make glider-toggleable data? Is it a boat-bit?

Another thing - Recently I read a post on this thread(?) about a still life that rotates itself 90º when hit by a glider. Are there any other known patterns that do this?

[simsim314]

Well, I just thought that the eater2 would be already known, but the other's wouldn't be.

It's very common theme in the gun collection. This is also very common problem in catalyst search (those are added to any escaping glider as valid catalysts). CatForce actually has special feature added to it to erase these cases (I think it's the only catalyst utility that fully solves this problem).

[gmc_nxtman]

Well, I just thought that the eater2 would be already known, but the other's wouldn't be.

It's very common theme in the gun collection. This is also very common problem in catalyst search (those are added to any escaping glider as valid catalysts). CatForce actually has special feature added to it to erase these cases (I think it's the only catalyst utility that fully solves this problem).

That's pretty cool!

Is there a search program for eaters? I saw a file in Golly's still-lifes folder with crazy novel eaters, but I don't know how to find them...

[simsim314]

Is there a search program for eaters?

Bellman with glider input, you'll find a lot but not all. In the collection of eaters, I guess someone just checked a lot of still life combinations.

[dvgrn]

Also, what is the smallest way to make glider-toggleable data? Is it a boat-bit?

It's hard to beat a boat-bit for size -- though it takes a bigger mechanism to read the state of the boat-bit, if you don't know whether it's there or not. Here's a recent example.

Another thing - Recently I read a post on this thread(?) about a still life that rotates itself 90º when hit by a glider. Are there any other known patterns that do this?

Couldn't find that recent post offhand -- I think I remember one where the still life is changed to its mirror image, but I'm not seeing that one right away either.

The only small one-glider 90-degree turn that I can recall uses a loaf target:

Code: Select all

x = 9, y = 7, rule = B3/S23
bo$2bo$3o$6bo$5bobo$5bo2bo$6b2o!

Loaves can also be flipped 180 degrees, or can be moved (1,3) with no rotation. I think that's not quite enough variety to put any orientation of loaf anywhere you want it, even allowing gliders from any direction. Unless you convert to some other object type and then back again, you're limited to one checkerboard square color (right?).

[gmc_nxtman]

Last but not least, this one generates "Trans-mirrored offset hooks" at about T=700 (that unfortunately gets destroyed) and later creates a cis-boat and dock:

Code: Select all

x = 9, y = 8, rule = Life
7bo$6b2o$7b2o3$b2o$obo$bo!

Is it known?

Code: Select all

x = 42, y = 9, rule = B3/S23
2obo2bo11b2ob2o10b2o5b2o$bob4o12bobo3bo7bobobo3bo$bo17bo2b4o9bob4o$2o
3b2o10bobo15bo$5bo11b2o5b2o8b2o3b2o$4obo17bobo13bo$o2bob2o10b4o2bo10b
4obo$17bo3bobo9bo3bobobo$20b2ob2o8b2o5b2o!

Also, are these loops known?

[simsim314]

, are these loops known?

Those are not loops, but C4 still lifes. And yes they are trivial.

I would suggest for starters to read some of the articles in the GOL links thread, to get more familiar with what's known and what's not.

[gmc_nxtman]

, are these loops known?

Those are not loops, but C4 still lifes. And yes they are trivial.

By loops, I meant mutually inducting C4 still lifes.

I would suggest for starters to read some of the articles in the GOL links thread, to get more familiar with what's known and what's not.

You're right, I will check those out.

[dvgrn]

I would suggest for starters to read some of the articles in the GOL links thread, to get more familiar with what's known and what's not.

Two more quick suggestions:

1) Check Mark Niemiec's new database. There's a known glider construction there for the four-tables still life.

2) look up objects by number of bits in Catagolue's search results. See for example the list of 32-bit still lifes for C4_1 symmetry, which includes the 32-bit C4 still lifes.

It's certainly not hard to create still lifes or find reactions that no one has ever seen before... the tough part is figuring out if whatever-it-is is unique enough to be worth advertising, or if it's just one of a large class of similar things that are all about as easy to find. If there are hundreds or thousands of equivalent items in the class, that tends to dilute the interest level a lot.

Catagolue has counted 1.8 million cis-rotated hooks and over a million cis boat and docks so far, so those are both fairly common objects, as these things go.

Is there a search program for eaters? I saw a file in Golly's still-lifes folder with crazy novel eaters, but I don't know how to find them...

Dean Hickerson found many of those with a search program called "dr", short for "drifter". Nowadays some of them could also be re-discovered with Mike Playle's Bellman program.

There are still plenty of new discoveries that could be made in this area -- new record longest recovery times, for example. And somewhere out there is the smallest "crazy novel eater" where the active spark finds its way to an edge of the eater and turns into a 90-degree glider, while the eater returns to its initial state. Unfortunately that kind of thing probably isn't findable by re-running searches using existing programs -- some kind of new approach will be needed.

[gmc_nxtman]

There are still plenty of new discoveries that could be made in this area -- new record longest recovery times, for example. And somewhere out there is the smallest "crazy novel eater" where the active spark finds its way to an edge of the eater and turns into a 90-degree glider, while the eater returns to its initial state. Unfortunately that kind of thing probably isn't findable by re-running searches using existing programs -- some kind of new approach will be needed.

Hmm- maybe people could submit still lifes to an online database, and have a program which just places a random glider to collide, and checks out what happens?

[simsim314]

Hmm- maybe people could submit still lifes to an online database

Why do you need online database for this?

and have a program which just places a random glider to collide, and checks out what happens?

This is exactly what bellman does, in extremely optimized way. To explain in short, take this reaction:

Code: Select all

x = 8, y = 9, rule = B3/S23
bo$2bo$3o3$4b2o$4bo$5b3o$7bo!

It's obviously an eater. Now there are many variations of the same basic interaction.

Code: Select all

x = 61, y = 13, rule = B3/S23
bo9bo9bo9bo9bo9bo$2bo9bo9bo9bo9bo9bo$3o7b3o7b3o7b3o7b3o7b3o5$6b2o8b2o
8b2o8b2o8b2o8b2o$6bo3bo5bo9bo9bo3bo5bo3bo5bo$7b4o6b3o7b3o7b4o6b4o6b3o$
20bo8bo30bo$9b2o6b2obo5bobo8b2o8b4o6b3o$9b2o6bobo6b2o9b2o8bo2bo6bo!

Bellman knows to automatically skip all the variations, once the pure core of the interaction was found.

More than that bellman knows to skip all failed cores that exploded as well.

[dvgrn]

Hmm- maybe people could submit still lifes to an online database, and have a program which just places a random glider to collide, and checks out what happens?

There are algorithms for randomly generating large still lifes, and it's easy to try all possible input gliders for a given still life. But unfortunately the Sun might burn out before the first randomly-generated still life happens to produce a glider and return to its initial state. There are just too many possible still lifes, and the odds are too low that a randomly chosen one will respond really nicely to being hit by a glider.

Of course we might get lucky... but it would be nice to improve the odds a little, above what we're likely to get from pure random trials.

There might be 10 trillion distinct 48-bit still lifes, and 25 trillion 49-bit still lifes, and 60-70 trillion 50-bit still lifes, and so on (assuming 2.5 times as many N+1-bit still lifes as N-bit still lifes, which seems to be about right.)

With modern computers it's easy to check a million different still lifes, and a billion is possible with enough time, and a trillion can maybe be done with enough time and a huge distributed search... but much above 50 bits and an exhaustive search starts to be completely out of reach, practically speaking.

H. Koenig at pentadecathlon.com has programmatically checked glider collisions with all still lifes up to some number of bits -- was working on 17-bit objects, I believe, but that was quite a few years ago. Very very occasionally a glider collision would increase the size of the still life, along these lines --

Code: Select all

x = 10, y = 7, rule = B3/S23
2b2obo2b2o$2bob2ob2o$9bo$6o$o4bo$3bo$3b2o!

-- but mostly the glider causes either a catastrophic explosion or (much more rarely) a partial collapse.

[Kazyan]

H. Koenig at pentadecathlon.com has programmatically checked glider collisions with all still lifes up to some number of bits -- was working on 17-bit objects, I believe, but that was quite a few years ago. Very very occasionally a glider collision would increase the size of the still life, along these lines --

Code: Select all

x = 10, y = 7, rule = B3/S23
2b2obo2b2o$2bob2ob2o$9bo$6o$o4bo$3bo$3b2o!

-- but mostly the glider causes either a catastrophic explosion or (much more rarely) a partial collapse.

Incidentally, is there a reference for collisions with Spartan objects? It's a simple thing, but I don't want to have to do that search again by hand.

[gmc_nxtman]

Hmm- maybe people could submit still lifes to an online database, and have a program which just places a random glider to collide, and checks out what happens?

There are algorithms for randomly generating large still lifes, and it's easy to try all possible input gliders for a given still life. But unfortunately the Sun might burn out before the first randomly-generated still life happens to produce a glider and return to its initial state. There are just too many possible still lifes, and the odds are too low that a randomly chosen one will respond really nicely to being hit by a glider.

Of course we might get lucky... but it would be nice to improve the odds a little, above what we're likely to get from pure random trials.

There might be 10 trillion distinct 48-bit still lifes, and 25 trillion 49-bit still lifes, and 60-70 trillion 50-bit still lifes, and so on (assuming 2.5 times as many N+1-bit still lifes as N-bit still lifes, which seems to be about right.)

With modern computers it's easy to check a million different still lifes, and a billion is possible with enough time, and a trillion can maybe be done with enough time and a huge distributed search... but much above 50 bits and an exhaustive search starts to be completely out of reach, practically speaking.

H. Koenig at pentadecathlon.com has programmatically checked glider collisions with all still lifes up to some number of bits -- was working on 17-bit objects, I believe, but that was quite a few years ago. Very very occasionally a glider collision would increase the size of the still life, along these lines --

Code: Select all

amazingness

-- but mostly the glider causes either a catastrophic explosion or (much more rarely) a partial collapse.

That collision....

Anyways, I think I've come up with a rule which someone could use in a catalyst search program to check for catalysts; in fact, one could have a catalyst search where you could choose for it to only show transparent results, or show both simply by having 2 rule files and depending on the option you choose, altering between the rule files.

The basic idea is that you have at least 4 states; lifecell, before-catalyst, after-catalyst, and dead cell.
When before-catalyst gets hit with a lifecell it turns into after-catalyst, and to only show transparent results the program would check if the entire catalyst turns into after-catalyst.

[Kazyan]

Check out the LifeHistory rule built into Golly; it works on the same principle as your rule idea. Looks like you're getting the hang of things work, but just aren't up to speed on all the minutae in this hobby...which is understandable, as we don't have a Conway's Game of Life Handbook, just a far-from-comprehensive wiki.

Actually...getting published would make things easier for my writing dream down the line...perhaps I should write one.

Anyway, if you liked that collision, this one upgrades a hook-with-claw into something with 6 additional cells.

Code: Select all

x = 7, y = 9, rule = B3/S23
2bo$obo$b2o3$2bobo$2b2obo$5bo$5b2o!

[Scorbie]

Anyway, if you liked that collision, this one upgrades a hook-with-claw into something with 6 additional cells.

I reckon that from the accidental discoveries thread, correct?

[gmc_nxtman]

Check out the LifeHistory rule built into Golly; it works on the same principle as your rule idea. Looks like you're getting the hang of things work, but just aren't up to speed on all the minutae in this hobby...which is understandable, as we don't have a Conway's Game of Life Handbook, just a far-from-comprehensive wiki.

Yeah... I just can't find that many interesting results by hand

Anyway, if you liked that collision, this one upgrades a hook-with-claw into something with 6 additional cells.

Code: Select all

x = 7, y = 9, rule = B3/S23
2bo$obo$b2o3$2bobo$2b2obo$5bo$5b2o!

That's pretty cool!

To make up for the lack of patterns in recent posts....

Code: Select all

#C [[ THUMBNAIL AUTOSTART LOOP 100 ZOOM 5 ]]
x = 13, y = 8, rule = Life
10bo$10b3o$10bobo$12bo$bo$obo$obo$bo!

Some sort of near miss for a blinker catalyst:

Code: Select all

x = 12, y = 4, rule = Life
9bo$9b3o$3o6bobo$11bo!

[simsim314]

There might be 10 trillion distinct 48-bit still lifes

You missing a point - eaters could consist of several SLs. Even if you take combinations of two SLs from 1M options you get 10^12, multiply it by location (say 10x10), and glider positioning (at least 10 as well) you easily get 10^15.

For example chris_c spartan eater that consists of 4 SLs requires the following search space:

- About 20 spartan SLs (counting each orientation individually). This comes four times.
- 10x10 search space. This calculated three times.
- 10 for glider position.
- divide by 4!

We get ~50 billions. Considering we need to run each combination and see the result for at least 10 iterations, 100K per second is pretty optimistic. So the time it will take is about 150 hours. And we're just talking about spartan SL space. With CatForce and more reasonable time estimation, such check would take few month.

----

On some other note: today to get billions you just need to use modern GPU. it's X10-20 times faster than the fastest multi-core cpu, and most of life search applications run on single core so this is about X100 acceleration.

Notice that many of the things we're talking about can run in parallel search pretty straight forward. I'm actually thinking about a way to use parallel computing for tree traversal as well, because it's so popular in GOL search utilities (gfind, bellman, catalyst to name a few).

[gmc_nxtman]

by simsim314 less than a minute ago

Dang, fast reply!

I'm currently running an apgsearch blonksoup thingy to search for eater2/ eater5 seeds...!

I hope I get some results.

By the way, nobody answered my question. Is the R_pentomino-boat interaction known?

(this was kinda rushed btw)

[Scorbie]

By the way, nobody answered my question. Is the R_pentomino-boat interaction known?

I am not sure what you mean by that, but if you meant this, then it is known.

viewtopic.php?f=2&t=576&p=13593&#p13593