Thread for basic questions

[dbell]

How should I construct a 28c/83 helix in Life(or PedestrianLife)?

Section 10.2.2 of the "Conway's Game Of Life - Mathematics and Construction" book by Nathaniel Johnston and Dave Greene describes a method to make a helix for any speed < c/2. (https://conwaylife.com/book/).

However, that method usually creates a large helix. To find a small helix is tricky. It seems that you have to scan through many combinations of several variables to find smaller solutions.

Finally, that method only uses three different reactions. Alternative reactions might work better.

BCNU,
-dbell

[WhiteHawk]

How should I construct a 28c/83 helix in Life(or PedestrianLife)?

Section 10.2.2 of the "Conway's Game Of Life - Mathematics and Construction" book by Nathaniel Johnston and Dave Greene describes a method to make a helix for any speed < c/2. (https://conwaylife.com/book/).

However, that method usually creates a large helix. To find a small helix is tricky. It seems that you have to scan through many combinations of several variables to find smaller solutions.

Finally, that method only uses three different reactions. Alternative reactions might work better.

BCNU,
-dbell

Also note that there are differences between life and Pedestrian Life in terms of how (classical) helices react/burn, which may be helpful

Turns out one of the ways that a glider can hit a xWSS diverges significantly from regular Life when B8 is added, and this difference allows for bizarre helix segments with a burning front faster than c/2 forwards. In this example (which may not be the most optimal cleanup) the burning front travels at (9,7)c/16:

Code: Select all

Just for demonstration, here's a 677c/1464 helix, which is faster than Life's speed limit of 11c/24 for classical helices. I have a strong suspicion that it's possible to push the asymptotic bound in B38/S23 to c/2 without having to rely on blinker fuses. EDIT: See attachment / note at the bottom.

Code: Select all

I'm probably not going to look into finding more compact ways to achieve this sort of thing right now, so it's open to anyone to try and improve on this (finding a better fuse slope with less severe geometric problems, less smoke, anything that makes high speed helices simpler)

EDIT (footnote): I realized immediately that I was right about c/2 being the speed limit here. Basically, a "good" slope can be obtained by just chaining mulitiple >c/2 speed chunks in a row with delays between them such that the burn speed averages out to exactly c/2. At this point, the chains go exactly orthogonally, and so an arbitrary amount of burn time can be fit into a constant height within the helix, thus allowing speeds arbitrarily closer to c/2 by just making the helix wider. I've attached an example that moves really fast just to prove this is the case. For many speeds the quadratic blinker fuse design is probably better as it needs far fewer ships and is more compact laterally, but this proves blinker fuses aren't strictly necessary here. The problem now is that of optimization - it's possible that more familiar helix speeds might be able to benefit if sufficiently good helix segments can be found.

[yyh_baboon]

Thanks!
From the book it seems that a 46c/166 helix can be constructed in just 9 WSS, which is pretty good

[parrot]

How do I turn a pattern into RLE without “fancy software”?

[b-engine]

How do I turn a pattern into RLE without “fancy software”?

If "fancy software" means CA simulator like Golly or LifeViewer, you can replace off cells with b, and on cells with o; use $ to proceed to the next line and ! to end the RLE:

Code: Select all

bo$
bbo$
ooo!

Still, using CA simulator is recommended here so others can easily view your pattern.

Also note that methuselah andstill life aren't generally notable enough to post here, if you don't know.

[parrot]

By “fancy software” I mean golly, and would you mind elaborating on the “LifeViewer” part? How do I use it to convert patterns to RLE?

[b-engine]

By “fancy software” I mean golly, and would you mind elaborating on the “LifeViewer” part? How do I use it to convert patterns to RLE?

LifeViewer can been accessed by clicking "SHOW IN VIEWER" of a code area.
It's also a CA simulator, so you can draw patterns, select using the select tool or Ctrl+A, then copy using the copy button or Ctrl+C. The copied pattern is the RLE which can been pasted anywhere.

You really shouldn't use any other means to save patterns.

[parrot]

Thanks! On another note, how do you get the trusted flag?

[I6_I6]

Thanks! On another note, how do you get the trusted flag?

Welcome to the forums!
You can request for it on the LifeWIki Trusted Account Request Thread:
viewtopic.php?f=16&t=2300

EDIT a couple days later:
How do I add pattern files to infoboxes in LifeWiki articles?

[b-engine]

Were there "fake patterns" that were considered groundbreaking but later found out to be fake (i.e. not behaving like it should be)?
Excluding those that are sufficiently small because everyone could simulate that and find out.

EDIT: removed the ' from were. that stupid autocorrect

[hotdogPi]

It wasn't groundbreaking (it was pretty much ignored), but there was a "fine structure constant calculator" that was just a reskin of the phi calculator that had the hex digits 0123456789APRILF (where it printed F00L).

viewtopic.php?p=42212#p42212

This is an exhaustive enumeration error rather than a specific pattern not working as intended, but I believe when the 20-cell still lifes were first enumerated way back, this one was missed.

Code: Select all

x = 11, y = 7, rule = B3/S23
5bo$4bobo$bo2bobo2bo$obobobobobo$bo2bobo2bo$4bobo$5bo!

[Chris857]

We're there "fake patterns" that were considered groundbreaking but later found out to be fake (i.e. not behaving like it should be)?
Excluding those that are sufficiently small because everyone could simulate that and find out.

It's small, but A for all had an invalid 13-glider synthesis from 1993/1995 that was not noticed to be invalid all the way until 2019.

[qqd]

Has 3-4 Life been proven to be Turing complete?
EDIT: Also, could someone provide the original source of the proof of the universality of the original sixteen Herschel conduits?

[dvgrn]

Were there "fake patterns" that were considered groundbreaking but later found out to be fake (i.e. not behaving like it should be)?
Excluding those that are sufficiently small because everyone could simulate that and find out.

Let's see, the first Spartan universal computer-constructor used a leaky mechanism for its memory tapes, with an escaping glider for every "1" stored. That meant it wasn't an entirely SUCCessful universal constructor -- e.g., it couldn't construct itself, it could only construct itself plus a bajillion gliders. It was a fixable problem, but a fix was never completed.

The longest-running fake pattern that I can think of was an April Fools joke that, as far as anyone ever told me, went unnoticed for more than seven years.. The fact that there's no possible algorithm for computing the fine structure constant should have tipped people off immediately, though. Quite possibly there were people who knew about the pattern and its claimed behavior, and knew that the claim couldn't be true, but just weren't interested in blowing the whistle on the joke.

[dvgrn]

EDIT: Also, could someone provide the original source of the proof of the universality of the original sixteen Herschel conduits?

"Universality" needs a footnote or two in this context -- it's something like

... with one or more additional mechanisms, oscillators of periods 61+n could be built!

That's from an 12 October 1996 re-posting to LifeCA of an essay by David Buckingham. There were eight "original Herschel conduits" before the "original sixteen", and Buckingham no doubt knew immediately that several three-conduit subsets were universal -- in the sense of 'you can build oscillators of any period above some minimum period P, using these three conduits".

Discovery order isn't entirely clear, so it's a bit of a puzzle what the first known "universal" subset of Herschel conduits was. It seems like that "universality" property was treated more like an obvious consequence of discovering enough stable Herschel conduits, rather than being formalized into a complete proof. Maybe that state of affairs lasted until Nathaniel did wrote up a proof for the Life textbook, but it's quite possible that somebody published a formal proof sooner than that.

[Sokwe]

could someone provide the original source of the proof of the universality of the original sixteen Herschel conduits?

"Universality" needs a footnote or two in this context -- it's something like

... with one or more additional mechanisms, oscillators of periods 61+n could be built!

That's from an 12 October 1996 re-posting to LifeCA of an essay by David Buckingham. There were eight "original Herschel conduits" before the "original sixteen", and Buckingham no doubt knew immediately that several three-conduit subsets were universal -- in the sense of 'you can build oscillators of any period above some minimum period P, using these three conduits".

Discovery order isn't entirely clear, so it's a bit of a puzzle what the first known "universal" subset of Herschel conduits was.

That article is the first published source of a set of conduits allowing the construction of oscillators of all sufficiently high periods (before that, only R64 had been published). The discovery dates for the different conduits are given in the article, so one can infer the timeline of oscillator period discoveries, which I did a few years back when trying to determine the first discovered oscillator for each period:

The following is a timeline of David Buckingham's discovery of all remaining periods above 60 using stable Herschel tracks:

• 5 Jul 1996: All periods 153 and above (using R64, L112, and F117) and all periods 64 and above that are not multiples of 3 or 13 (using just L112 and F117)
• 7 Jul 1996: All periods 107 and above (using L112, F117, and R190)
• 3 Aug 1996: All periods 61 and above (using L112, F117, and Fx77)

[I6_I6]

What's the purpose of friends and foes?

Also, in LifeWiki, how do you add pattern files & glider syntheses? I want to add them to the articles for both unidimensional spaceships.

[b-engine]

What's the purpose of friends and foes?

These are the built-in features of phpBB, which is the forum board software that this forum uses.

[rabbit]

If you add someone as a foe, it prevents their posts from appearing in threads locally. I'm not sure what adding someone as a friend does besides put them on a quick access list on the PM page.

[dvgrn]

Also, in LifeWiki, how do you add pattern files & glider syntheses? I want to add them to the articles for both unidimensional spaceships.

In general, if the pattern is too big to create a pname and an "RLE:pname" page but small enough that it's good to have it in the all.zip pattern collection, you can post files (or links to existing file locations) in the Suggested LifeWiki edits thread.

If RLE:pname is created, it will semi-automatically get uploaded to the all.zip patterns collection at the beginning of every month. But don't do that for multi-megabyte RLEs. It would bulk up all.zip too much if we started trying to throw huge spaceships or huge synthesis files into it. Usually it's good enough to add a link in the LifeWiki article to the forum post where the pattern is attached.

Also, I'm still looking for Pattern Collector volunteers who would be interested in working on modernizing Golly's "Very Large Patterns" online archive -- that collection is really far behind the times at this point, and this unidimensional spaceship stuff could definitely fit there.

[I6_I6]

How do inflated symmetries on Catagolue work?
I understand what inflation is, I just don't understand the symmetry names, like iiC1 or iiii8x32.
Also, what's the purpose of inflated symmetries?

[PK22]

How do inflated symmetries on Catagolue work?
I understand what inflation is, I just don't understand the symmetry names, like iiC1 or iiii8x32.
Also, what's the purpose of inflated symmetries?

Each 'i' in the symmetry's prefix means one iteration of inflation is applied to the soup. iiC1 is a C1 soup inflated twice, so each 1x1 cell becomes 4x4, and iiii8x32 is a 8x32 soup inflated four times, so each 1x1 cell becomes 16x16.

The most useful purpose is for finding Margolus oscillators in rules that support them - this was one of the intentions behind the original idea when it was implemented, but really they can just be used as larger soups if that turns out to be useful in some situations.

[I6_I6]

Each 'i' in the symmetry's prefix means one iteration of inflation is applied to the soup. iiC1 is a C1 soup inflated twice, so each 1x1 cell becomes 4x4, and iiii8x32 is a 8x32 soup inflated four times, so each 1x1 cell becomes 16x16.

Then why are the soups in iiii8x32 16x16? I understand that the soups are non-inflated versions, but then shouldn't the soups for iiii8x32 (and others) be 8x32?

[PK22]

Each 'i' in the symmetry's prefix means one iteration of inflation is applied to the soup. iiC1 is a C1 soup inflated twice, so each 1x1 cell becomes 4x4, and iiii8x32 is a 8x32 soup inflated four times, so each 1x1 cell becomes 16x16.

Then why are the soups in iiii8x32 16x16? I understand that the soups are non-inflated versions, but then shouldn't the soups for iiii8x32 (and others) be 8x32?

If Catagolue's soup generator (found in the repo at src/main/java/com/cp4space/catagolue/census/Soup.java) doesn't know how to generate a soup, it will just return the C1 soup for the given hash, which is why it is returning C1 soups.

[I6_I6]

This used to be in the thread for advanced questions, but I moved it here (and deleted the original):

...
EDIT:
Is it possible for there to be an oscillator in a square grid alternating rule whose mod is not equal to, half of or a quarter of the full period?
Or can one just have multiple mods that aren't even a factor of the full period?

EDIT 2:
My second question has been proven true.
I'm sure there's a simpler solution, but here is a p4 which flips itself in 3 generations, then flips itself back in 1:

Code: Select all

x = 2, y = 3, rule = B3ae/S1c2c3e|B2e3e/S2a3ei
o$bo$o!

What would be the mod of such an oscillator?