Clearing pattern

[sipos]

This is a newbie question: I haven't thought much about Conway's Life before. I'd appreciate an answer though if someone doesn't mind answering.

Is there any pattern that persists and moves, like a glider, but which clears the board as it moves?

[Schiaparelliorbust]

This is a newbie question: I haven't thought much about Conway's Life before. I'd appreciate an answer though if someone doesn't mind answering.

Is there any pattern that persists and moves, like a glider, but which clears the board as it moves?

What do you mean by "clear the board"? It's by default already empty.

[sipos]

Sorry, I should have been clearer.

What I mean by clear is "kill" cells that have been left alive by other patterns operating, so in the path of this pattern would only be dead (unset) cells, and whatever patterns it encountered in front of it (I guess with some exceptions) it would continue moving, leaving no alive cells behind it. Basically, like a glider, but that erases (most) things in-front of it.

[Schiaparelliorbust]

Sorry, I should have been clearer.

What I mean by clear is "kill" cells that have been left alive by other patterns operating, so in the path of this pattern would only be dead (unset) cells, and whatever patterns it encountered in front of it (I guess with some exceptions) it would continue moving, leaving no alive cells behind it. Basically, like a glider, but that erases (most) things in-front of it.

There are sparky spaceships and you can also use rakes to shoot gliders at objects but that would probably create more junk. There is Bulldozer, which can delete tiny objects in front of it with its strong front spark, hence the name.

Code: Select all

x = 47, y = 76, rule = B3/S23
20$19b3o$20b3o5$19b2o$19b2o5$18b3o8$15bo6bo$14bobob2obobo$14b2o2b2o2b
2o$18b2o2$16b2o2b2o$16b2o2b2o$17b4o$17bo2bo$16bo4bo$16bo4bo$16bob2obo
$17bo2bo$17bo2bo2$11bo14bo$11b2o4bo2bo4b2o$11bo5b4o5bo$12b2o4b2o4b2o$
13bo10bo$11bobo10bobo$12b5o4b5o$14bo2b4o2bo$13bo4b2o4bo$17bo2bo$18b2o
$15b2ob2ob2o2$14bo8bo$13b3o6b3o$13bo2bo4bo2bo$12b2o10b2o!

In the future, please ask questions like these on the thread for basic questions instead of creating a new thread.

[sipos]

Thanks. I'll check those out.

Sorry, I should have usefd the basic question thread. TBH, I am sufficiently new to this to not even know how basic a question is, but I should have defaulted to that.

[dvgrn]

Thanks. I'll check those out.

Sorry, I should have usefd the basic question thread. TBH, I am sufficiently new to this to not even know how basic a question is, but I should have defaulted to that.

Not to worry, we can handle a new thread every now and then. Welcome to the forums!

Another way of approaching an answer is to say that if we know exactly where every ON cell is, "out there" in the universe, then we can probably design a pattern that cleans it up by carefully shooting gliders at it, or building spark-making objects that do the cleanup work, etc.

Even if we know there's some incredible huge castle with complex defenses in the path we want to clear, something as big as the Death Star, we can probably figure out how to fire a few gliders at it and blow it up, and then clean up the ash once it settles down -- as long was we have the complete blueprint, so that we can calculate exactly what the ash will settle into.

However, if we don't know the positions of all cells in advance, then we don't have a good way in Conway's Life of doing "remote sensing" for arbitrary objects. We could build something that works to clean up blocks, for example, or even sense the presence or absence of blocks and build one or more copies of the block field while destroying it. But if something other than a block shows up, we're just as likely to get an explosion that makes the cleanup problem much worse, than to get a successful cleanup.

Most people don't think there is likely to be a mechanism that can "reach out" into space full of random ash, and reliably clean it out to make empty space. Certainly I personally would much rather have the task of finding an arrangement of objects that breaks a candidate "space cleaner", than the task of designing and building such a candidate!

[dvgrn]

Here's something I wrote in an email to simsim314, last time this topic came up:

> I was thinking how we can define CGOL as containing creatures that will control their environment out of infinite random soup....
> Do you think such task is even doable in cgol? Do we have such "probes" of a sort?

A GPC hooked up to a universal constructor is certainly capable of
setting up a specific "cleaning" reaction at each cell in an
arbitrarily large range, say a glider hitting a sacrificial block like
this to clear out blocks if they exist at the very edge of the current
range:

Code: Select all

x = 52, y = 9, rule = LifeHistory
4.2A18.2A18.2A$4.2A18.2A18.2A4$11.2C38.C$3A8.2C7.3A7.3C7.3A8.C$2.A19.
A19.A8.C$.A19.A19.A!

This one works for one phase of blinker (creating a block slightly
farther away which will be cleaned up in a later cycle) -- but not the
other phase, so we'd need something better than this. Also of course
if the first blinker phase is too close to some other still life, or
if it's part of a traffic light, or if the test reaction hits a
honeyfarm instead, all bets are off.

Basically, if Conway's Life were one of those rules where there are
only a couple of kinds of still life, plus a few really really rare
things, then we might get away with building some kind of guard fence,
reaching out through it with a construction arm, cleaning out a large
space by testing one cell at a time, and maybe returning a "success"
signal after each time that nothing unexpected happened. Whenever a
signal fails to come back, you start cleaning the entire space again.

That way if you hit one of the really rare things that your cleanup
reaction wasn't designed to handle, the odds are that you'll turn it
into a bunch of common stuff, and the cleaning will be successful the
next time through... except you might want to tear down and rebuild
the guard fence before every attempt, to be safe.

But in B3/S23, there are thousands of different objects that you're
bound to come across while trying to clean out random ash, if you
include pseudo-objects and various common closely-spaced
constellations that might get sparked into an explosion when one of
their components gets "cleaned". The explosions can be a fairly large
step backward in the cleaning process, and so far no one has come up
with a specific cleaning reaction that, on average, reliably reduces
the population of Unknown Ash Space.

Part of the problem is that we don't have a really reliable guard
fence, either. The best known arrangement is a thick layer of widely
spaced blocks, just far enough apart that a pi explosion from one
block doesn't touch any other blocks, and thick enough that there are
no open paths all the way through the blocks. That way, at least
singleton gliders will never create new gliders inside the guard fence
and create a chain reaction that way.

But unexpected explosions in Unknown Ash Space very regularly create
two gliders in the same direction, and quite often they make *WSSes,
and sometimes they make block-laying or glider-producing switch
engines... and again all bets are off. It's very difficult to find a
mechanism that can reliably recover if it's cleaning out a 100Kx100K
space, and happens to spark a GPSE aimed unluckily right at the
mechanism, starting from the far edge of that space.

A shorter answer: this topic has come up every half-decade or so, on
the LifeCA mailing list and on the conwaylife.com forums, but so far
no specific "cleaner-on-average" mechanisms have been suggested, Given
the existence of GPSEs and BLSEs, especially, the odds don't seem
terribly good that any such thing exists. At the very least, they're
apparently hard to find!

[MathAndCode]

However, if we don't know the positions of all cells in advance, then we don't have a good way in Conway's Life of doing "remote sensing" for arbitrary objects. We could build something that works to clean up blocks, for example, or even sense the presence or absence of blocks and build one or more copies of the block field while destroying it. But if something other than a block shows up, we're just as likely to get an explosion that makes the cleanup problem much worse, than to get a successful cleanup.

Most people don't think there is likely to be a mechanism that can "reach out" into space full of random ash, and reliably clean it out to make empty space. Certainly I personally would much rather have the task of finding an arrangement of objects that breaks a candidate "space cleaner", than the task of designing and building such a candidate!

My best idea for accomplishing this task is making the "reaching out" mechanism as unlikely to spark chaos as possible by giving the region as few additional cells with which to create chaos. Gliders are a good candidate, but I'm thinking that we should try edgeshooting either blocks or blinkers, and I'm leaning towards blocks because they tend to have at least two cells die whenever they interact with something and because oscillators are slightly closer to chaos than still lifes. The method that I'm thinking of would be something like this, where the destroyer edgeshoots a block cell by cell and then row by row, except valid.

Code: Select all

x = 108, y = 63, rule = TripleB3S23
107.G$105.2G$2.G103.2G$G.G$.2G19$73.G.G$73.2G$74.G4$47.G13.G$48.G11.G$46.3G11.3G4$50.G$50.G.G4.G$50.2G3.G.G$56.2G$45.G$46.2G14.G$45.2G13.2G$61.2G18$58.CA$58.CA$58.B!

[toroidalet]

The problem is that there's always an option that results in a ton of junk, although if you really wanted to you could probably make a blonk-clearer.

Code: Select all

x = 108, y = 64, rule = B3/S23
107bo$105b2o$2bo103b2o$obo$b2o19$73bobo$73b2o$74bo4$47bo13bo$48bo11bo$
46b3o11b3o4$50bo$50bobo4bo$50b2o3bobo$56b2o$45bo$46b2o14bo$45b2o13b2o$
61b2o18$58b2o$57bo2bo$57bo2bo$58b2o!

Yes, you could change the reaction slightly so it killed ponds too, but you'd never be able to cleanly destroy a giant still life, like an extremely long snake with a big blob at the end custom-designed to explode when given the clean fuse your reaching out method generates.

[MathAndCode]

The problem is that there's always an option that results in a ton of junk, although if you really wanted to you could probably make a blonk-clearer.
Yes, you could change the reaction slightly so it killed ponds too, but you'd never be able to cleanly destroy a giant still life, like an extremely long snake with a big blob at the end custom-designed to explode when given the clean fuse your reaching out method generates.

Yes, but as long as we have some mechanism to prevent chaos from spreading too much and some mechanism to tell the destroyer when chaos has formed, we only need to be able to handle the ten or so most common objects, as the destroyer can simply destroy the ash of any rare object. Unfortunately, this particular method does not work well with beehives in a certain orientation.

Code: Select all

x = 108, y = 64, rule = B3/S23
107bo$105b2o$2bo103b2o$obo$b2o19$73bobo$73b2o$74bo4$47bo13bo$48bo11bo$46b3o11b3o4$50bo$50bobo4bo$50b2o3bobo$56b2o$45bo$46b2o14bo$45b2o13b2o$61b2o18$58bo$57bobo$57bobo$58bo!

[toroidalet]

The problem is that there is no way to detect the formation of chaos without potentially creating more (to do so, you would have to hit the area where there may or may not be junk with gliders or something, which might cause another explosion).

Another problem is that if an explosion is created, it could send gliders or LWSSes or even switch engines back towards your mechanism, which would have to be cleaned up without knowing where or when they will strike (see above; also note that any potential cleanup salvo would need to block such entities in all phases and locations, including from the sides or even from behind).

Also, as dvgrn mentioned earlier, it's pretty hard to find a reaction that makes the ash frontier retreat on average because of the ridiculously large number of different ways to trigger explosions.


In short, Life is just too chaotic for a universal clearer to exist, but it might be possible if we assume that the ash is only composed of a small group of objects (which negates all the problems by making it possible to have a clean destruction reaction).

[hkoenig]

You might also want to consider the that at a high level this question was posed in LifeLine #7 as "The Irresistible Force Problem: ... This is easily proved impossible -- just shoot two at each other."

[Moosey]

You might also want to consider the that at a high level this question was posed in LifeLine #7 as "The Irresistible Force Problem: ... This is easily proved impossible -- just shoot two at each other."

^

Meanwhile we still haven't proven or disproven the existence of an indestructible static object-- a "universal eater"