Starting Out

[QED]

I'm rather new to CGoL, and I'd like to start learning about how to actually help out, like engineering things instead of blowing up the Turing Machine in Golly. However, there is no actual tutorial/guide to newcomers basically ANYWHERE. I'm not sure where to begin, or where to start. Basically, my only experience with cellular automata is passively studying Rule 110. Can anyone help?

[biggiemac]

Welcome!

This forum points to a lot of good resources, see the "Game of Life Links" thread. Quite a few more good resources are in the other threads but it's a lot to wade through and the search function is a bit lacking. I think there's a post by dvgrn somewhere listing many different directions someone interested in CGOL construction could go.

Unfortunately there can't be a single tutorial to cover it all, since different types of projects require entirely different ways of thinking and methods of attack, develop their own vocabulary, etc.. In many cases programs are written tailored to the project, so a lot of larger-scale projects benefit greatly from the ability to write and understand Python, (or Lua, which is now a supported programming language in Golly). It is by no means a requirement though!

Typically familiarizing yourself with what has been created (and the tools used) is a best first step to understanding how to create things of your own.

So, given there are a lot of different directions you could go, are there particular projects that interest you?
Creating fast engineered spaceships like Caterpillar or Waterbear? There's an unfinished (27,1)c/72 spaceship that I've been away from for a while but was my last significant project.
Slower, methodical self-replicating designs? There is still no complete quadratic replicator in CGoL, but a number of people have been throwing around mechanisms for engineering one.
Producing guns to fire spaceships of unique speeds? That usually comes not too long after a glider synthesis is complete - the topic of glider syntheses has a lot of focus lately.
Improving "glider circuitry" methods? In some cases that requires running programs that search for small reactions, but in other cases one can do some clever engineering with parts already known. I think the "construction practice" thread has very recently been optimizing a glider-to-c3 converter this way.
Universal spaceship designs for adjustable high speed? simsim314's caterloopillar can be generalized to speeds higher than c/4 but it's not a simple problem and while there have been a few posts giving proofs of concept it still has a few obstacles someone can work on solutions to. (Though it's possible he has done more work he hasn't shared)

There's tons more, but those listed are the most engineering-centered ones that come to mind.

[QED]

All this looks very cool. I think I'd like to work on spaceship desgin, self-replication, and guns. Glider circutry looks cool, but sounds very complex. So, how can I start going about actually working on these things? I'm not sure where to start, or how to start designing.

[dvgrn]

All this looks very cool. I think I'd like to work on spaceship desgin, self-replication, and guns. Glider circutry looks cool, but sounds very complex. So, how can I start going about actually working on these things? I'm not sure where to start, or how to start designing.

Okay, here's something completely random, and maybe a little bit ambitious, but it's a place to start:

A new cheaper glider recipe for a 3-engine Cordership showed up just yesterday, so it would be perfectly reasonable -- as these things go, anyway -- to build a gun using the new recipe. Maybe post it, or incremental steps toward it, on the Construction Practice thread.

If you really want to jump in at the deep end, you could work on a glider-to-Cordership converter instead of just a regular gun. With one of those, you can attach a single gun of any high-enough period -- or any kind of signal stream you want, doesn't have to be regular -- and, hey presto, Corderships come out in the same pattern. If you want a different period, you can just swap out the one small drive gun.

For that you might try out this glider adjustment toolkit (still in draft form) and tell me on that thread what you find horribly confusing about all the long boring details that I've written, and I can try to improve and explain.

I have some notes which will eventually turn into a glider-to-Cordership V-converter (gliders coming from just two banks of edge-shooters, as in the last recipe in the first link above)... if nobody else gets there first, anyway. I was thinking maybe I'd write up the construction process, for either that converter or a new V-shaped G-to-loafer converter, as a walkthrough -- would that be interesting at all? The loafer only needs eight or ten synchronized gliders, instead of 11 or 13 or 15 for the Cordership. So maybe that would be a slightly better starting point for you.

EDIT: For any project along these lines you'll need edge shooters. There's a collection higher up in the Construction Practice thread, followed by lots of examples of how they're used.

[QED]

Wow... that's insanity. I looked at the gun for the 15- glider version and it's crazy. Is there any easier first project?

[dvgrn]

Wow... that's insanity. I looked at the gun for the 15- glider version and it's crazy. Is there any easier first project?

Of course! But it partly depends on which gun you looked at. Was it the p690 version? You have to learn a fair amount to be able to sort out all the moving p30-based parts.

A stable 14-glider Cordership maker, with recipe gliders coming from three sides, is a lot simpler in some ways. There aren't any moving parts -- the pattern just sits there and waits for a glider to arrive and start it up. So you can build it a small piece at a time without worrying that the parts will get out of sync with each other.

Now, glider-to-Cordership converters used to be really crazy, too, but there have been a number of recent discoveries that make things a whole lot easier, and smaller.

Just a thought: one possible easier project is to build just a third of the Cordership converter at a time, starting with the easiest third first. If anyone else wants to learn some of this Life circuitry stuff, maybe a collaboration could get going.

For the northeast side you'd start with one glider, and connect it to a H-to-2G converter and a few reflectors, to get two gliders out in the correct places relative to each other. That's pretty much the smallest possible synchronization problem, and it can be done with trial and error.

Then for the southwest side it's about twice as hard -- four synchronized gliders.

And if you decide to tackle it, the northwest side is twice as hard again, eight gliders. Makes kind of a gradual progression.

---------------------------------

But okay, I won't push for my particular random idea any more than that. There are lots of other things you could experiment with.

Another idea that came up was to look for the smallest possible mechanism to edge-shoot a fishhook eater. Let's say a stream of gliders is going by, and then this mechanism off to one side gets triggered to build an eater right on the glider lane. Then all future gliders in the stream get absorbed by the newly built eater.

It only takes two gliders to build an eater, and technically we do have things kind of like this already, in universal regulators and elsewhere. But there's probably a smaller faster easier way to do it, using recently discovered H-to-2G components.

Here's one more random idea: I suppose this is also a little crazy to look at, but you could try adjusting Jormungant's telegraph to print a different message. By the time you play around with it for a while, figure out which parts you should throw away and which you should keep, you'll probably understand a lot more about Life circuitry.

The telegraph encodes letters in binary as patterns of 5 glider locations, with the glider either there (1) or not there (0). A space is 00000, an A is 00001, and so on. So if you find the right clues in the thread, it's really just a matter of adding or deleting a few gliders to change which character gets displayed.

[Goldtiger997]

Wow... that's insanity. I looked at the gun for the 15- glider version and it's crazy. Is there any easier first project?

These days, most discoveries made, are done by scripts. However, there are exceptions, such as construction and glider synthesis.

Here's a first project for you, find a glider synthesis for 16.142 in under 16 gliders. This is the still life 16.142:

Code: Select all

x = 5, y = 7, rule = B3/S23
2bobo$bob2o$bo$2ob2o$bobo$o2bo$b2o!

I quickly looked at catalogue, and found a simple predecessor for it. Here it is with the syntheses of all the separate parts included in the pattern:

Code: Select all

x = 102, y = 42, rule = B3/S23
36bob2o$36b3o6b2o$37bo7b2o$33bo$33bo5bo$33bo4bobo$39bo23$97bo$32bo65bo
$30bobo63b3o$obo28b2o67bo$b2o96bo$bo29bo42bo24b3o$31b2o34bobo3b2o$b2o
27bobo34b2o4bobo$2b2o64bo$bo$68b2o$68bobo$68bo!

What you need to do is to make all of the syntheses of the separate parts happen at the right time, so that they match the predecessor at the top, which will make 16.142. If you complete it you can post it in the 16 in 16 thread. If it's too hard, I can give more help.

[muzik]

This thread details a few things that need done, but they aren't quite beginner things I would say.

[dvgrn]

Wow... that's insanity. I looked at the gun for the 15- glider version and it's crazy. Is there any easier first project?

...
Another idea that came up was to look for the smallest possible mechanism to edge-shoot a fishhook eater. Let's say a stream of gliders is going by, and then this mechanism off to one side gets triggered to build an eater right on the glider lane. Then all future gliders in the stream get absorbed by the newly built eater.

It only takes two gliders to build an eater, and technically we do have things kind of like this already, in universal regulators and elsewhere. But there's probably a smaller faster easier way to do it, using recently discovered H-to-2G components.

Actually this might be a better idea than I thought. The stable universal regulator just came up on the "Things to be discovered" thread, and I went and had a look at it. It needs a drive gun with a period of 1177 ticks or more, and it kinda sorta edge-shoots a glider -- at least, it would if you turned it inside out, routing both of the synchronized rebuild signals clockwise.

Whoever built that universal regulator clearly had absolutely no idea what they were doing.** There's a whole extra completely unnecessary G-to-H-to-G component in each of the two signal branches, that really doesn't do anything but waste time.

It should be possible to replace the H-to-2G in the center of the universal regulator. Pick the correct H-to-2G from the draft glider adjustment toolkit, and use only Snarks for the rest of the reflectors. It should be possible to cut the bounding box considerably, and get the minimum drive gun period well down below 1000, at the very least.

This is another case where the synchronization problem is as simple as it can possibly be -- only two gliders. And it's a useful mechanism, so either

1) people would actually copy and use it for years in other patterns, and the pattern would end up in the LifeWiki with your name on it;

or

2) your improvement would make other people pay attention to the problem, and one or more of them would find even smaller faster solutions, and then the smallest fastest one would end up in the LifeWiki instead.


** [Full disclosure: it was me, four years ago.]