Thread for basic questions

[Vort]

what's the smallest number of logic gates needed to implement cgol step?

I remember some gpu implementation doing amazing stuff, but my search came up only with this 2009 page with 35 gates
...

Here is what written on Lifelib page:
... Conway's Game of Life can be computed in 19 Boolean operations ...

[Citation needed]

Is there a list of known "XWSS-to-XWSS" turners (including gliderless ones)?

[Hippo.69]

It's real in the sense that it's a video of a real pattern being run in Golly with the help of a real script. Pattern and script can be found here.

It's also real in the sense that, even though that sample "semilator" starting pattern consists of more than fifteen gliders, it's just fifteen gliders plus a bunch of "demonstration" xWSSes. That starting pattern is easy to convert into an equivalent but much larger fifteen-glider starting pattern that constructs the exact same thing (but that is much too large for Golly to run successfully).

That's why I did see a horizontal line besides the 15 gliders in the video. That's why I asked the question.

What did the xWSSes do?
(It's still unbelieveable for me that 4 GPSE plus some extra gliders can construct everything constructible.)

Also, the synthesis for GPSE needs glider that comes opposite to the GPSE's direction -- I wondered if you can make sure it's a valid synth. (If you have noticed that, you will use the kickback reaction, won't you? but it costs extra gliders.)

EDIT: I see. So in the demo pattern, most of work are done by the xWSSes, but there are some work by the GPSEs which shows the actual thing. I know how it works without the xWSSes.

EDIT 2: Can you provide a pure RCT15 pattern that generates a shillelagh?
EDIT 3: Is there any script that converts any valid synthesis to a RCT15 pattern?

This is slightly weird, but has anyone ever submitted RCT-based syntheses to Catagolue?

Original version of RCT15 was highly impractical with about 1500000 bits required to construct an empty pattern.
(Aditional bits are required to build the pattern). I have tried a lot (a year of optimizations) to make it less highly impractical and now about 600000 bits are required to construct the empty pattern (and the encodding of pattern construction is bitwise a bit more efficient).
This means the multiplicative constant in the number encoding the pattern roughly got third root.
Number having 600000 binary digits (encoding the distance from the epicenter) is big enough to be practical.
If you have a display where a pixel measures the plank length of the size of observable universe, it will have about (10^{60}~2^{200}) pixels in one dimension.
Showing RCT15 pattern on such a display would be problematic, as you would require it 2^{-599800} zoomed out.
This is good incentive to show the pattern on logarithmically smaller scale ... and use xWSSes to insert the bits.
28 bits are encoded in the distance in the demo patterns.

Curently we have 3 RCT15 demos ... first building pattern issuing digits 01234567890123456... based on LWSS's

... the 1500000 version,
pattern seed of an empty space

... the 600000 version and pattern generating a spacefiller

... 600000 version (+ about 280000 more bits to encode the pattern).

I am working on 4th demo ... pattern generating digilat clock. ... the computation is going about 1 year now with few (about a weak) interruptions.
Still decomposing the pattern seed to a glider salvo. (I have made some minor changes to pslmake so now it does not need an assistance ... it is almost done now.) Next step will be optimization of the salvo to fire gliders in close lanes.
Then the lane program (the salvo) would be inserted to the lane program of the ECCA arm of the RCT15 and new bits will be recomputed.
Estimate of the bits encoding the pattern is in order of 10^7. I am afraid the demo would require more than 28 bits to be encoded by the distance as the corderships cleaning the GPSE remnants could reach the GPSE starting locations too early so the glider invoking the seed would arrive before the seed would be constructed.
This is not problem in full RCT15 version as the distance to the starting locations would be sufficiently large .

You could consider shillelagh to be an empty pattern in this scale especially as with the empty pattern you have to add an object to anihilate the incomming glider to make a shillelagh seed.

BTW: the optimized version of RCT15 project had to take the demo version into account ... few additional bits could be saved, but the xWSSes would collide with the construction (the method how GPSE's are stopped and cleanup of the epicenre is done ... crossing the XWSS's line during the stoppers construction).

Considering the script ... you have to provide the SW salvo constructing a seed to be ignited by a SE traveling glider (with a spare additional object to anihilate from the SW traveling glider) ... Usually the spare additional object could be find by special "agnosticize of a salvo" ... finding alternative lanes of the same construction ... (it saves a lot of bits in DBCA programs, but only few in ECCA programs).

I should specify the relative position of the starting target block of the construction salvo relative to the SE traveling glider. ... I will EDIT the post ...

Then all ECCA salvos (in lane phase format) are connected by a .bat (in my case) to a huge lane phase file, which is compiled by ECCA bits compiler.
The bits generated by the complier are joined with bits of all previous phases of RCT15 project and semilate program is run which translates it to the pattern consisting of gliders and xWSSes (and thanks to comments in the bit files semilate could generate rct15_show.lua with the pattern file as well).

[dl-rs]

When I was chatting with abother CA enthusiast one day, I asked him if the conjecture 'no infinite growth pattern in CGOL below 10 cells' has been proven for any 10 cell pattern, or just patterns in a certain bounding box, because if we can place a pre-pre-behive, a pre-block, a blinker, etc. far away from an R, maybe we can also get a large chaotic pattern tha evolves into a BLSE or GPSE, but I'm not sure if this has been tried. Has it?

[dvgrn]

When I was chatting with abother CA enthusiast one day, I asked him if the conjecture 'no infinite growth pattern in CGOL below 10 cells' has been proven for any 10 cell pattern, or just patterns in a certain bounding box, because if we can place a pre-pre-behive, a pre-block, a blinker, etc. far away from an R, maybe we can also get a large chaotic pattern tha evolves into a BLSE or GPSE, but I'm not sure if this has been tried. Has it?

The Infinite growth pattern article is pretty clear on this point --

Nick Gotts and Paul Callahan have since shown that there is no infinite growth pattern with fewer than 10 cells ...

-- but unfortunately no references are provided. Should probably fix that.

Checking old Life email, I can find

From: Paul Callahan
Date: Tue, 30 Dec 1997 15:30:50 GMT
Subject: 10-cell single-cluster infinite growth pattern


I ran an enumeration of 10-cell single clusters, which is about the limit
of the method I'm using on the machines I'm using. It's too slow to
simulate all the clusters, so I kept only those that have 11 or
more cells for each of the 10 generations after the initial pattern, and
rule out duplicates before checking for infinite growth. This ran for
several days and used a significant percentage of available memory because
of the need to rule out duplicates. (luckily I was able to keep it running
over the holidays when the machine would have been idle for the most part).
This led to roughly 600000 patterns that needed to be simulated; I have
all of these in compressed form, so I can run some other tests on them
if I can think of any. Going through this set takes much less time than
generating it did.

Anyway, I found one infinite growth pattern. It produces a block-laying
switch-engine:

*.....
......
**....
......
..***.
......
...***
....*.

There may be be other 10-cell patterns that lead to the same final state,
because of the way I ruled out patterns.

I also looked for patterns that produce two or more spaceships (but found
none that produced more than two). The most interesting such pattern
I found produces a symmetric pair of LWSSs as well as a symmetric
induction-coil still life (itself the result of a symmetric pair of
B-heptominoes).

*.........
.*****.***
*.........

I don't expect there to be any 9-cell infinite growth patterns. I have
run through what I think are all cases with the (now correct) quiescence
detector. I still have to go back and document the cases I checked to
make sure I didn't miss any.

--Paul

From: Paul Callahan
Date: Sat, 3 Jan 1998 11:03:46 -0500 (EST)
Subject: Re: 10-cell single-cluster infinite growth pattern


H. Koenig writes:

>Is it safe to assume that a side effect of these searches is that it can
>safely be concluded that there aren't any previously undiscovered
>long-period oscillators or spaceships of [10 or fewer cells]?

I was going to answer no to your question, since my old program would have
failed to find a new period-60 oscillator. However, this one should have
reported any subpattern oscillating at a higher period than 15. In
principle, my program could be ignoring unusual low-period spaceship (i.e.,
testing them for collisions but not noticing that they are special), but
I don't think this is very likely.

--Paul

But I'm having a hard time digging up Nick Gotts' contribution, which I think was the enumeration of all of the ways that two small blobs totalling 9 cells could interact with each other. Does anyone have a reference for that? I seem to remember it having happened, but I don't remember where it was documented.

[Sokwe]

When I was chatting with abother CA enthusiast one day, I asked him if the conjecture 'no infinite growth pattern in CGOL below 10 cells' has been proven for any 10 cell pattern, or just patterns in a certain bounding box, because if we can place a pre-pre-behive, a pre-block, a blinker, etc. far away from an R, maybe we can also get a large chaotic pattern tha evolves into a BLSE or GPSE, but I'm not sure if this has been tried. Has it?

The Infinite growth pattern article is pretty clear on this point --

Nick Gotts and Paul Callahan have since shown that there is no infinite growth pattern with fewer than 10 cells ...

-- but unfortunately no references are provided. Should probably fix that.

Nick Gotts' fairly detailed analysis can be found in this thread. If you read the thread carefully, you will see that he accounts for things like an R-pentomino emitting a glider that collides with a distant object. The announcement of the result that there are no infinite growth patterns with fewer than 10 cells is given in this post.

[islptng]

How to synthesize a spaceship from ikpx2 or other spaceship-searching programs?

[PK22]

How to synthesize a spaceship from ikpx2 or other spaceship-searching programs?

Synthesising any large spaceship that can only be found by search programs like ikpx2 can be extremely difficult - the Space dust page likens the task to "building a Formula-1 race car whilst it's on the track and travelling at 200mph", as you need a way to get every cell in the correct position with the correct timing. It has been possible in some cases, however, such as with the Spider (although it originally costed 601G, which has since been reduced to 201G).

Typically, large spaceship syntheses are created when someone finds a 'final step' that produces the spaceship, which will be a constellation of still lifes and/or oscillators being hit by gliders and/or xWSSes. Then, others will try to find a way to synthesise the constellation used in the final step, and the synthesis is complete.

I am curious though - how do people typically find these final steps? Is it done with a search program?

[ec1263f3]

[...] I am curious though - how do people typically find these final steps? Is it done with a search program?

I suggest to check the first five footnotes in Tutorials/Glider syntheses, as well as the discussion in viewtopic.php?f=2&t=335&start=175 with replies by vilc and Goldtiger997.

[WhiteHawk]

Here is the activation step for a p19 honey farm hassler

This is an interesting variant of Goban. Notably, while it has some similarities with the more standard Goban variant, it does have some differences (see below for example). What is this variant's discovery information?

Code: Select all

x = 86, y = 110, rule = B3/S238
4b2o$4b2o38$10b2o$10bobo$13bo2b2o$11b2obo2bo$10bobob2o49b2o9b2o$11bo53b
o9bobo$21b2o44bo2b2o2bo$21bo44b5o2bob2o$6bo5b3o4bobo33bo7b2o7b2obobo$
5bobo3bo3bo3b2o4bo29b3o5bob4o7bo$5b2o3bo5bo7bobo31bo10bo$3b2o5bo5bo7b
obo30b2o5b2o2b2o$2bo2b2o3bo5bo6b2ob3o35b2o7b3o5bo2b2o$bobo2bo4bo3bo13b
o21b2o19bo3bo3bobo2bo$o2b2o7b3o8b2ob3o22bo7b3o9bo5bo3b2obo$2o21b2obo26b
o4bo3bo8bo5bo5bo$6bob2o21b2o19b2o3bo5bo7bo5bo3b2o$4b3ob2o8b3o7b2o2bo18b
o5bo5bo8bo3bo4bo$3bo13bo3bo4bo2bobo18bob2o3bo5bo9b3o7bo$4b3ob2o6bo5bo
3b2o2bo18bo2bobo3bo3bo19b2o$6bobo7bo5bo5b2o19b2o2bo5b3o7b2o$6bobo7bo5b
o3b2o37b2o2b2o5b2o$7bo4b2o3bo3bo3bobo37bo10bo$11bobo4b3o5bo31bo7b4obo
5b3o$11bo45bobob2o7b2o7bo$10b2o46b2obo2b5o$21bo38bo2b2o2bo$17b2obobo34b
obo9bo$15bo2bob2o35b2o9b2o$15b2o2bo$20bobo$21b2o38$27b2o$27b2o!

[Chris857]

viewtopic.php?f=2&t=1437&p=178897&hilit=p19#p178961
It was posted here by Period1GliderGun shortly after goban was found

[dl-rs]

It is impossible that this reaction/conduit was unknown, so what is the name of this and where is it used in?

Code: Select all

x = 8, y = 13, rule = B3/S23
o$2o$b2o$bo$o3$6b2o$3bo3bo$3b4o2$3b2o$3b2o!

[dvgrn]

It is impossible that this reaction/conduit was unknown, so what is the name of this and where is it used in?

Code: Select all

x = 8, y = 13, rule = B3/S23
o$2o$b2o$bo$o3$6b2o$3bo3bo$3b4o2$3b2o$3b2o!

That's an HF2P. (It doesn't start with a B-heptomino, because then a block gets left behind and it's not a valid conduit.)

[Quantum-mechanics]

How to measure offset of conduit.
Edit:

[...]
Use the diagram at the bottom of this post.

Each type of active object shown in the diagram -- Herschel, B-heptomino, R-pentomino, etc.) has one particular cell highlighted as its "zero point" -- the white cell.

The offset of a conduit is the (X,Y) distance from the white cell location in the conduit's input, to the white cell location in the conduit's output.

For each signal type in the diagram, the location of the white cell is somewhat arbitrarily chosen; statistics could be collected equally well if different "zero point" locations were chosen, as long as the choice is consistent. The above linked post records the arbitrary decisions we've made about those locations, so that we can collect consistent information about elementary conduit offsets.

Like? PT8P?

[dvgrn]

How to measure offset of conduit.

Use the diagram at the bottom of this post.

Each type of active object shown in the diagram -- Herschel, B-heptomino, R-pentomino, etc.) has one particular cell highlighted as its "zero point" -- the white cell.

The offset of a conduit is the (X,Y) distance from the white cell location in the conduit's input, to the white cell location in the conduit's output.

For each signal type in the diagram, the location of the white cell is somewhat arbitrarily chosen; statistics could be collected equally well if different "zero point" locations were chosen, as long as the choice is consistent. The above linked post records the arbitrary decisions we've made about those locations, so that we can collect consistent information about elementary conduit offsets.

[Selena Silverstep]

Do we know the fate of all patterns fitting in a 5x5 bounding box? What about larger sizes?

[PK22]

Do we know the fate of all patterns fitting in a 5x5 bounding box? What about larger sizes?

Here is a Catagolue census of all patterns in a 5x5 bounding box. There are 8 block laying switch engines, for example, although they all probably come from different rotations and reflections of the same 5x5 soup.

We probably know/can search larger sizes - you can use gsearch, which can exhaustively search patterns inside a given bounding box, with 4x7 taking about a day. However, I think that the time taken for a search could grow faster than 2^(m*n), where m and n are the dimensions of the bounding box, and at some point, completing a very large gsearch search with current technology is not feasible in the lifetime of the universe.

Edit: Whoops, corrected gfind to gsearch. gfind is a program for finding spaceships.

Here's a calculation (mainly for fun) showing that you will also be limited by energy:
500 watts (power consumption of a computer running at full power) * 86400s (seconds in a day) = 4.32*10^7J of energy per day.
3.86 x 10^26 watts (power output of the sun) * 5 billion years (estimated remaining lifetime) = 6.0906168 × 10^43J output by the sun in its remaining lifetime.
2^28 soups (number possible in a day) / (4.32*10^7J) = 6.2137837 soups/J.
6.2137837 soups/J * 6.0906168*10^43J = 3.7845775*10^44 soups.
Bear in mind that this would require humanity to start capturing all of the sun's energy, starting right now, but by the end, there is only enough energy to search a 74*74 bounding box, and it would take an insane amount of time to search.

Other issues would probably crop up - cosmic rays will interfere, the computer's hardware would erode, and I am not sure if gsearch can handle quadratic growth patterns. The smallest known quadratic growth pattern (by bounding box) fits into a 97 × 33 bounding box, so one in a 74x74 box would be likely to appear. There is also a very slim chance that something similar to IceNine could appear and cause massive slowdowns.

[Resu]

When was this Z-hexomino agar first discovered?

Code: Select all

x = 59, y = 75, rule = B3/S23
32b2o$33bo$30b2obo$31bob2o$28b2obo$29bob2ob2o$26b2obo5bo$27bob2ob2obo
$24b2obo5bob2ob2o$25bob2ob2obo5bo$22b2obo5bob2ob2obo$23bob2ob2obo5bob
2o$20b2obo5bob2ob2obo$21bob2ob2obo5bob2ob2o$18b2obo5bob2ob2obo5bo$19b
ob2ob2obo5bob2ob2obo$16b2obo5bob2ob2obo5bob2o$17bob2ob2obo5bob2ob2obo
$14b2obo5bob2ob2obo5bob2ob2o$15bob2ob2obo5bob2ob2obo5bo$12b2obo5bob2o
b2obo5bob2ob2obo$13bob2ob2obo5bob2ob2obo5bob2ob2o$10b2obo5bob2ob2obo5b
ob2ob2obo5bo$11bob2ob2obo5bob2ob2obo5bob2ob2obo$8b2obo5bob2ob2obo5bob
2ob2obo5bob2o$9bob2ob2obo5bob2ob2obo5bob2ob2obo$6b2obo5bob2ob2obo5bob
2ob2obo5bob2ob2o$7bob2ob2obo5bob2ob2obo5bob2ob2obo5bo$4b2obo5bob2ob2o
bo5bob2ob2obo5bob2ob2obo$5bob2ob2obo5bob2ob2obo5bob2ob2obo5bob2o$2b2o
bo5bob2ob2obo5bob2ob2obo5bob2ob2obo$3bob2ob2obo5bob2ob2obo5bob2ob2obo
5bob2ob2o$2obo5bob2ob2obo5bob2ob2obo5bob2ob2obo5bo$bob2ob2obo5bob2ob2o
bo5bob2ob2obo5bob2ob2obo$bo5bob2ob2obo5bob2ob2obo5bob2ob2obo5bob2ob2o
$b2ob2obo5bob2ob2obo5bob2ob2obo5bob2ob2obo5bo$5bob2ob2obo5bob2ob2obo5b
ob2ob2obo5bob2ob2obo$2b2obo5bob2ob2obo5bob2ob2obo5bob2ob2obo5bob2o$3b
ob2ob2obo5bob2ob2obo5bob2ob2obo5bob2ob2obo$3bo5bob2ob2obo5bob2ob2obo5b
ob2ob2obo5bob2ob2o$3b2ob2obo5bob2ob2obo5bob2ob2obo5bob2ob2obo5bo$7bob
2ob2obo5bob2ob2obo5bob2ob2obo5bob2ob2obo$4b2obo5bob2ob2obo5bob2ob2obo
5bob2ob2obo5bob2o$5bob2ob2obo5bob2ob2obo5bob2ob2obo5bob2ob2obo$5bo5bo
b2ob2obo5bob2ob2obo5bob2ob2obo5bob2o$5b2ob2obo5bob2ob2obo5bob2ob2obo5b
ob2ob2obo$9bob2ob2obo5bob2ob2obo5bob2ob2obo5bob2o$9bo5bob2ob2obo5bob2o
b2obo5bob2ob2obo$9b2ob2obo5bob2ob2obo5bob2ob2obo5bob2o$13bob2ob2obo5b
ob2ob2obo5bob2ob2obo$10b2obo5bob2ob2obo5bob2ob2obo5bob2o$11bob2ob2obo
5bob2ob2obo5bob2ob2obo$11bo5bob2ob2obo5bob2ob2obo5bob2o$11b2ob2obo5bo
b2ob2obo5bob2ob2obo$15bob2ob2obo5bob2ob2obo5bob2o$12b2obo5bob2ob2obo5b
ob2ob2obo$13bob2ob2obo5bob2ob2obo5bob2o$13bo5bob2ob2obo5bob2ob2obo$13b
2ob2obo5bob2ob2obo5bob2o$17bob2ob2obo5bob2ob2obo$17bo5bob2ob2obo5bob2o
$17b2ob2obo5bob2ob2obo$21bob2ob2obo5bob2o$18b2obo5bob2ob2obo$19bob2ob
2obo5bob2o$19bo5bob2ob2obo$19b2ob2obo5bob2o$23bob2ob2obo$20b2obo5bob2o
$21bob2ob2obo$21bo5bob2o$21b2ob2obo$25bob2o$25bo$25b2o!

[islptng]

I failed to install QuFince on my computer. Ubuntu@WSL@Windows 11.

Code: Select all

islptng@LAPTOP-C4S2ULR2:~$ git clone https://gitlab.com/apgoucher/lifelib.git
Cloning into 'lifelib'...
fatal: unable to access 'https://gitlab.com/apgoucher/lifelib.git/': gnutls_handshake() failed: Error in the pull function.

I downloaded the zip file, unzipped it, and opened it:

Code: Select all

islptng@LAPTOP-C4S2ULR2:~/lifelib$ ./compile_qufince.sh
-bash: ./compile_qufince.sh: Permission denied
islptng@LAPTOP-C4S2ULR2:~/lifelib$ sudo ./compile_qufince.sh
[sudo] password for islptng:
sudo: ./compile_qufince.sh: command not found

[Resu]

What is the corderman's unknown deconimo?

[b-engine]

What is the corderman's unknown deconimo?

The decomino was lost in record, therefore the exact pattern isn't known.

[hotdogPi]

What is the corderman's unknown deconimo?

The decomino was lost in record, therefore the exact pattern isn't known.

No, we've checked all of them. There isn't one.

[Resu]

The decomino was lost in record, therefore the exact pattern isn't known.

No, we've checked all of them. There isn't one.

Where was it first mentioned?
What is the smallest P-12 bi-block puffer?

[PK22]

Where was it first mentioned?
What is the smallest P-12 bi-block puffer?

1). The switch engine was first mentioned in Lifeline Volume 4, but where the statement about a decomino predecessor originates from is unclear - the wiki says 'Records indicate that', but no citation is provided for the records. Some more info can be found in this post.
2). I think that it is this combination of 2 backrakes:

Code: Select all

x = 41, y = 54, rule = B3/S23
2bo7bo19bo7bo$b3o5b3o4b3o3b3o4b3o5b3o$2obo4b2obo4bo2bobo2bo4bob2o4bob
2o$3o13bo7bo13b3o$b2o13bo7bo13b2o$7b2o8bobobobo8b2o$5b2o2bo21bo2b2o$
10bob3o11b3obo$6bo2bo21bo2bo6$12bo15bo$13bo13bo$11b3o13b3o7$15bo9bo$
16bo7bo$14b3o7b3o7$18bo3bo$19bobo$17b3ob3o6$18b2ob2o$18b2ob2o5$18b2ob
2o$18b2ob2o5$18b2ob2o$18b2ob2o!

[Resu]

2). I think that it is this combination of 2 backrakes:

Code: Select all

x = 41, y = 54, rule = B3/S23
2bo7bo19bo7bo$b3o5b3o4b3o3b3o4b3o5b3o$2obo4b2obo4bo2bobo2bo4bob2o4bob
2o$3o13bo7bo13b3o$b2o13bo7bo13b2o$7b2o8bobobobo8b2o$5b2o2bo21bo2b2o$
10bob3o11b3obo$6bo2bo21bo2bo6$12bo15bo$13bo13bo$11b3o13b3o7$15bo9bo$
16bo7bo$14b3o7b3o7$18bo3bo$19bobo$17b3ob3o6$18b2ob2o$18b2ob2o5$18b2ob
2o$18b2ob2o5$18b2ob2o$18b2ob2o!

What is the smallest c/2 bi-block puffer that is this period?

Code: Select all

x = 22, y = 5, rule = B3/S23
2o2b2o2b2o2b2o2b2o2b2o$2o2b2o2b2o2b2o2b2o2b2o2$2o2b2o2b2o2b2o2b2o2b2o
$2o2b2o2b2o2b2o2b2o2b2o!