Thread for basic questions

[b-engine]

Are x and y in RLE useless?

Yes, and in fact you can delete the x & y and replacing rule with xrule while having the entire RLE working:

Code: Select all

xrule = B3/S23
2o$b2o$bo!

[dvgrn]

Are x and y in RLE useless?

Mostly but not entirely. Try pasting some RLE like this into Golly:

Code: Select all

x = 99, y = 99, rule = B3/S23
49$49bo!

There are cases where you want to paste a selection with a carefully chosen offset and overall size -- and you can't do that without the x = , y = RLE header.

The "x" at the beginning of the non-comment section of the RLE is also a very useful hint for parsers that RLE is coming.

Yes, and in fact you can delete the x & y and replacing rule with xrule while having the entire RLE working...

I'd advise against posting patterns in b-engine's abbreviated form, at least for the moment. The minimal header would be

x=0,y=0,rule={rule}

Golly currently errors out in a hugely ugly way with an "xrule" pattern, and even the new safeopenclip.lua can't figure out what to do. Only LifeViewer is currently tolerant enough to support such a monstrously abbreviated format.

EDIT 6/12/2024: I've gone ahead and done a global search-and-replace on "xrule", replacing it with "x=0,y=0,rule" wherever that was appropriate -- so all of those patterns can now be copied into Golly without getting a twenty-line error message.

More to come about this. With any luck a better alternative can be found for posting (especially) experimental empty LifeViewers with various rules, which I think people are probably using almost exclusively via smartphones. It's certainly true that standard RLE format is awkward to type out on a phone -- probably awkward enough to cause some frustration and make the "xrule" shortcut look really good by comparison.

[wwei47]

Does anyone have slow salvo syntheses for forward LWSSes? I completely bypassed the glider-pair rake issue and am now bottlenecked by LWSS slow salvo track size. I am happy to try pretty much anything reasonable, since glider cost isn't anywhere near perfectly correlated with track space.

[dvgrn]

Does anyone have slow salvo syntheses for forward LWSSes? I completely bypassed the glider-pair rake issue and am now bottlenecked by LWSS slow salvo track size. I am happy to try pretty much anything reasonable, since glider cost isn't anywhere near perfectly correlated with track space.

What about the ones in the post by calcyman that you pointed to? It would be entertaining to be able to use the beehive-with-tail one. Is it too wide, or too slow?

I still haven't dug deep enough into the actual problem to be clear about whether difficulties arise when the slow salvo lanes are too wide, or maybe if it's the LWSS-producing reaction being too wide that's the problem, or something else. If it's the former -- do you know how to ask slsparse to build limited-width slow salvos for known forward LWSS seeds like the old Blockic one by knightlife?

[wwei47]

Does anyone have slow salvo syntheses for forward LWSSes? I completely bypassed the glider-pair rake issue and am now bottlenecked by LWSS slow salvo track size. I am happy to try pretty much anything reasonable, since glider cost isn't anywhere near perfectly correlated with track space.

What about the ones in the post by calcyman that you pointed to? It would be entertaining to be able to use the beehive-with-tail one. Is it too wide, or too slow?

Tried that one; too much track space as things currently are.

I still haven't dug deep enough into the actual problem to be clear about whether difficulties arise when the slow salvo lanes are too wide, or maybe if it's the LWSS-producing reaction being too wide that's the problem, or something else. If it's the former -- do you know how to ask slsparse to build limited-width slow salvos for known forward LWSS seeds like the old Blockic one by knightlife?

Right now it's literally just a track space issue.

[dvgrn]

Right now it's literally just a track space issue.

What is your definition of "track space"? Does that mean that the total width of the lanes taken up by the slow salvo?

If so, do you know how to ask slsparse to build limited-width slow salvos? And if you do, what forward-LWSS seeds have you tried compiling, at what lane widths?

Unfortunately there are a lot of possible width-N options to try -- you can slide the "allowable zone" in one direction or the other and end up with a different search with potentially very different results.

[wwei47]

Right now it's literally just a track space issue.

What is your definition of "track space"? Does that mean that the total width of the lanes taken up by the slow salvo?

It's how much space the rakes to make the slow salvo would take up on the actual block tracks.

If so, do you know how to ask slsparse to build limited-width slow salvos? And if you do, what forward-LWSS seeds have you tried compiling, at what lane widths?

Unfortunately there are a lot of possible width-N options to try -- you can slide the "allowable zone" in one direction or the other and end up with a different search with potentially very different results.

The last time I tried to get slsparse to work ended in failure; I don't know what to do.

[Drelectron8]

Are there any other self-complementary rules apart from just Day & Night (B3678/S34678) and it's strobing equivalent B01245/S0125?

If so, how do you find them?

[dvgrn]

The last time I tried to get slsparse to work ended in failure; I don't know what to do.

Can you work through the slsparse tutorial, and tell me what step you get to before something goes wrong? "Ended in failure" doesn't give me much to work with.

The width-limitation functionality is only mentioned in the first item in the "More Features" section. I can try running a few slsparse searches at various widths, and see what kinds of recipe lengths come back. Will probably put the best result into the tutorial as an example.

[dvgrn]

Are there any other self-complementary rules apart from just Day & Night (B3678/S34678) and it's strobing equivalent B01245/S0125?

If so, how do you find them?

There are 512 self-complementary outer-totalistic rules, if you count strobing rules, or 256 otherwise. See the Self-complementary article, and maybe also the Black-white reversal article:

Each rule has precisely one black/white reversal; if this is the same as the rule itself, the rule is said to be self-complementary.
...
To determine the black/white reversal of a given rule:

Negate the rule's B and S conditions, yielding B′ and S′.
Subtract each condition in B′ and S′ from 8, yielding B″ and S″.
The black/white reversal of B/S is S″/B″.

And there's a worked example after that. Basically from the "B" part of any rulestring, you can generate the "S" part that will make it self-complementary -- or equivalently generate the "B" part from any "S" part.

[wwei47]

The last time I tried to get slsparse to work ended in failure; I don't know what to do.

Can you work through the slsparse tutorial, and tell me what step you get to before something goes wrong? "Ended in failure" doesn't give me much to work with.

For some reason, it works perfectly fine now, despite me being unable to even clone the repository before.

The width-limitation functionality is only mentioned in the first item in the "More Features" section. I can try running a few slsparse searches at various widths, and see what kinds of recipe lengths come back. Will probably put the best result into the tutorial as an example.

I should explain how track space works shouldn't I.

Rakes shoot gliders on every 27th lane. Observe that rakes and related objects ("elements") will shift the track 4 lanes downwards. This means that by adding a rephaser between two rakes, the lanes that the second rake would shoot on shift downwards 4 lanes, or equivalently, up by 23 lanes. This means that the particular lanes on which gliders are shot are far more important than the total number of gliders.

For example, the second glider here is four lanes below the first one, meaning that we can perform this slow salvo using only two elements, both of which would be rakes.

Code: Select all

x = 34, y = 33, rule = B3/S23
2o3b2o$2o3bobo$5bo28$31b2o$31bobo$31bo!

On the other hand, the second glider here is on the same lane as the first one. Since the first lane will shift the track 4 lanes downwards, we will need 26 rephasers to shift the second rake so that it shoots on the same lane as the first rake, meaning that 28 elements total will be required. This slow salvo ends up taking up far more track space than the previous slow salvo mentioned, despite the fact that both slow salvos only have two gliders.

Code: Select all

x = 38, y = 33, rule = B3/S23
2o3b2o$2o3bobo$5bo28$35b2o$35bobo$35bo!

In general, if the next glider is n lanes below the current glider, you will need to add 7(n-4) mod 27 extra rephasers to get the next rake to shoot the next glider on the correct lane. This also makes me wonder if slsparse is the correct program for this, because slsparse does not understand that the number of gliders in the slow salvo is not nearly as important as the lanes on which those gliders are shot. Additionally, it seems like you have to specify a starting location for the block in slsparse, which is not something I care about; if the LWSS is in the "wrong" place, I should be able to line up other elements to make it "right".

Edit: You can just write 7n-1 mod 7 [EDIT 3: 27 not 7] instead of 7(n-4) mod 27.
Edit 2: Just tried running slsparse on this:

Code: Select all

x = 24, y = 28, rule = B3/S23
6b2o$6b2o2$2o$2o10$8b2o$8b2o8$21bo$17b2obobo$17b2obo2bo$21bobo$22bo!

After manual optimization to remove the extra initial push and stuff like that, I have this.

Code: Select all

x = 1797, y = 1793, rule = B3/S23
4b2o$2ob2o$2o3bo125$134b2o$133b2o$135bo126$261b3o$261bo$262bo126$395b
3o$395bo$396bo126$518b2o$517b2o$519bo127$632b2o$631b2o$633bo126$761b2o
$761bobo$761bo126$886b2o$886bobo$886bo126$1012b2o$1011b2o$1013bo126$
1138b2o$1138bobo$1138bo126$1267b2o$1266b2o$1268bo126$1391b2o$1390b2o$
1392bo126$1530b2o$1530bobo$1530bo252$1794b3o$1794bo$1795bo!

Apparently, this would take 183 elements to complete plus 6 elements to make the blocks and then line up the initial glider, which um...would take up even more track space. The cool beehive-with-tail seed you mentioned can be made to take only 71 elements in comparison plus those 6 elements.

[dvgrn]

In general, if the next glider is n lanes below the current glider, you will need to add 7(n-4) mod 27 extra rephasers to get the next rake to shoot the next glider on the correct lane. This also makes me wonder if slsparse is the correct program for this, because slsparse does not understand that the number of gliders in the slow salvo is not nearly as important as the lanes on which those gliders are shot.

Yup, this is all very familiar territory from the chapter about the Silverfish from the Life textbook. It's true that you can't really conveniently get slsparse to give a genuinely optimized answer to this kind of thing -- you can only pick the cheapest option out of whatever options that it presents... but if that cheapest option is anywhere near the actual best choice, it will only be by sheer luck.

What you really want is some custom code that builds an exhaustive search tree of slow salvo recipes, listing the salvos that can be made with each number of rephasers and rakes -- 1R, 2R, 3R ... until the first forward LWSS recipes start appearing.

If I remember right, that kind of search was done for the old 31c/240 Centipede. I think there was a more recent project that did something similar, but can't bring it to mind immediately. Can anyone point to any likely search code that might be adapted to run a search along these lines?

[wwei47]

it will only be by sheer luck.

Well, asking slsparse to build various seeds has gotten the cost down to 62 elements, but I doubt that it's enough:

Code: Select all

x = 1031, y = 1028, rule = B3/S23
4b2o$2ob2o$2o3bo127$136b2o$135b2o$137bo125$266b2o$265b2o$267bo127$390b
2o$389b2o$391bo125$519b3o$519bo$520bo127$648b2o$647b2o$649bo126$782b2o
$781b2o$783bo125$896b2o$895b2o$897bo127$1029bo$1028b2o$1028bobo!

[dvgrn]

it will only be by sheer luck.

Well, asking slsparse to build various seeds has gotten the cost down to 62 elements, but I doubt that it's enough...

Yup, that looks like it's in the ballpark for that method of finding recipes. LWSS seeds start showing up in the 7 / 8 / 9 glider range -- but if we enumerated recipes in order of cost, we'd probably find a first forward LWSS seed that was a lot cheaper to build than that.

[wwei47]

Yup, that looks like it's in the ballpark for that method of finding recipes. LWSS seeds start showing up in the 7 / 8 / 9 glider range -- but if we enumerated recipes in order of cost, we'd probably find a first forward LWSS seed that was a lot cheaper to build than that.

Hm...how would we write such a search program?
EDIT: I broke the quote oops
Edit 2: It turns out that the 62R I found won't actually work due to the LWSS being created on a row with the wrong parity, so it won't actually help here.
Edit 3: I'm just going to hack apgsearch since I don't know how to actually enumerate recipes in order of cost. This produces a forward LWSS and a ridiculous amount of junk!

Code: Select all

x = 820, y = 811, rule = B3/S23
4b2o$2ob2o$2o3bo66$68b2o$67b2o$69bo198$283b2o$282b2o$284bo66$346b3o$
346bo$347bo66$410b3o$410bo$411bo66$474b3o$474bo$475bo66$538b3o$538bo$
539bo66$603b2o$602b2o$604bo62$689b3o$689bo$690bo66$753b3o$753bo$754bo
66$818b2o$817b2o$819bo!

[miha1]

Code: Select all

x = 7, y = 3, rule = B3/S23
4b3o$3o$4b3o!

what is this? i cant identify it.

[JP21]

Code: Select all

x = 7, y = 3, rule = B3/S23
4b3o$3o$4b3o!

what is this? i cant identify it.

that is a different variant predecessor of the thunderbird
comparison:

Code: Select all

x = 52, y = 3, rule = B3/S23
4b3o40bo$3o44bob3o$4b3o40bo!

[Drelectron8]

Is there a pattern of all the discovered oscillators in CGOL? I would like to know.

[dvgrn]

Is there a pattern of all the discovered oscillators in CGOL? I would like to know.

Not quite -- or rather, not nearly! -- but there have been a couple of curated collections. Dean Hickerson put together a stamp collection of oscillators that were known by 1995 -- not a complete collection even then, of course, because it has been easy from very early on to invent arbitrary numbers of new oscillators by expanding old ones. (The Java doesn't work on those "Expand" links any more, but the RLE pattern links can be copied into Golly or LifeViewer, or the bitmaps can be pasted into Golly directly.)

Then hotdogPi did a lot of work much more recently to expand the oscillator stamp collection with LifeViewer labels -- though I somehow didn't manage to get the latest version from October 22, 2023 checked in to the recent Golly 4.3 release.

[dvgrn]

Edit 3: I'm just going to hack apgsearch since I don't know how to actually enumerate recipes in order of cost. This produces a forward LWSS and a ridiculous amount of junk!

Currently working on the enumeration problem on a thread in Discord. Why is the forward-LWSS-with-explosive-junk relevant?

Do you have examples of a rakes-and-rephasers design producing a slow salvo LWSS recipe along the lines of the 62R one that doesn't work? It seems like these recipes might also need to be edge-shooting, which (if it's true) is a key additional constraint that hasn't been mentioned yet.

[wwei47]

Edit 3: I'm just going to hack apgsearch since I don't know how to actually enumerate recipes in order of cost. This produces a forward LWSS and a ridiculous amount of junk!

Currently working on the enumeration problem on a thread in Discord. Why is the forward-LWSS-with-explosive-junk relevant?

I meant it jokingly; I was hoping that it would at least prove the concept, but then every other LWSS result turned out to be salvos that stopped being slow because I have to put the gliders too close together because putting them farther away would cause apgsearch to falsely detect these "soups" as stabilizing.

Do you have examples of a rakes-and-rephasers design producing a slow salvo LWSS recipe along the lines of the 62R one that doesn't work? It seems like these recipes might also need to be edge-shooting, which (if it's true) is a key additional constraint that hasn't been mentioned yet.

See the files attached here. viewtopic.php?f=2&t=2286&start=100#p183913

The recipe itself doesn't need to edgeshoot; the 2G LWSS turner will handle any necessary edgeshooting.

[Mathemagician314]

Are there any three-cell spaceships in outer-totalistic CA? If none are known, is it possible to prove that there are none?

[confocaloid]

Are there any three-cell spaceships in outer-totalistic CA? If none are known, is it possible to prove that there are none?

This spaceship has three alive cells, and these rules are outer-totalistic:

Code: Select all

x = 2, y = 3, rule = example
.B$A$.B!

@RULE example

@TABLE
n_states:3
neighborhood:vonNeumann
symmetries:permute

0, 0,0,0,1, 1
0, 0,0,1,2, 2
1, 0,0,0,0, 0
2, 0,0,0,0, 0

[Mathemagician314]

Sorry - I meant life-like CA.

[MasterbuilderMs]

I want to build some kind of wall, that if a glider hits it at any point, it will be destroyed. Fish hooks work, but only if the glider comes in the precise direction and position. Is this possible, or can I arrange fish-hooks in a specific way to cover every position?

EDIT by dvgrn: I'll move this post to the Basic Questions thread -- after waiting a day or two, just to make sure the question and answer don't get lost.