Langton's ant | Hacker News Reader

Langton's ant

en.wikipedia.org

118 points_piyush42 comments

I've been obsessed with Langton's ant for a good solid decade now. My latest creation is a few years old now, but occasionally I go back and add new commands. It's inspired by Langton's ant, but it operates in 45 degree increments, has the ability to fork, conditionally execute instructions, has colors, and a bunch of other interesting things. You can see it in action at http://demoseen.com/langton/#.FP$!!!!!!!!!!!!!!!!!!!!~ and there's a simple instruction listing at the bottom of the page.

It's quite a departure from the original, but you can make absolutely gorgeous images with some simple instructions.

This is awesome - lot of fun to play with.

This is the best pattern I've been able to come up with so far:

http://demoseen.com/langton/#..rr@<>@!!@@

That's really cool -- I love 'lattice' type patterns. This is my current favorite, because it seems to terminate (stabilize) and then does one last little bit. Runs for a good while before it 'terminates', as well. http://demoseen.com/langton/#.$!~PF!!!!.

After playing with this a good bit, here's my favorite:

http://demoseen.com/langton/#@@1o..+..-1o..+..-<>..$.^

It makes a cloth/napkin, and then slowly absorbs blood from the four corners.

That's really, really cool. I want to make a gallery system for these, where people can submit theirs and others can modify from there. Btw, I just added a polar mode (the grid is treated as polar coordinates, and converted from there); the & instruction switches to polar and _ brings it back to cartesian, so you can actually blend the two. http://demoseen.com/langton/#&m.$!~PFP!!..

Wow! Very cool... I could play with this for days! Thank you for posting this.

This is amazing work. Langton's ant was one of the first programs I wrote in Java around 1996 so it's super fun to see what you've done with it.

Wow! A suggestion: Adding some sort of inertia calculation. It might lead to very complex and yet smooth looking images.

Beautiful. Thanks!

any way to simulate turmites with this?

But what happened to Langton himself? After leaving SFI and the Swarm corp he seems to have disappeared. I met him once - about 12 years ago - and he was the most charming, kind guy who spent three hours and a lunch with me just because I said I don't understand something he said. Then he warned me not to become too much of a generalist and specialise more. He was right, but I didn't listen to him, I was in my 20s and thought I can be a polymath :)

I spoke to Glen Ropella of Swarm about him in 2009 when I was writing a paper and Glen said this: "Most of the people I know related to Swarm refer people to me when they're looking for Chris. It's no mystery, though. He decided awhile back to leave the Swarm/SFI community for personal reasons. So, he's out there. He just doesn't want to work on that stuff anymore, at least as of the last time I talked to him. ... He may well want to interact. I don't know. All I know is I haven't heard from him in awhile."

Chris, if you're out there: Hi!

I looked for Langton myself years ago when I was very into artificial life and evolutionary computing. IMHO he's one of the greatest unsung geniuses of the past 25 years.

Here's what I consider his greatest work:

http://wiki-app2.tudelft.nl/pub/Education/SPM955xABMofCAS/Le...

To give you one example: that paper is why I think Titan is the most likely world in the solar system where we might find present-day complex life. (Other than Earth of course.)

Why? There's no liquid water on the surface, and it's too cold!

Answer: phase boundaries everywhere. See Fig. 3 from paper.

Titan has a hydrocarbon-based "water cycle" with solid, liquid, and gas all existing at the same time, along with what appear to be seasons. Universal computation occurs in the vicinity of phase boundaries, so I would not be terribly surprised if we found "cryolife" there. It would be radically different from our own, but not really... I agree with people like Langton that life should be considered primarily an informatics phenomenon rather than a conventional chemical reaction or physical state.

If Titan life were intelligent, they'd regard us as hell-beasts with blood of molten water. :) We could never physically touch, as we would vaporize them and they would freeze us solid.

There's a contemporary scientist named Dr. Christoph Adami at Michigan State's new BEACON center who's probably the closest to following in Langton's footsteps.

http://www.mmg.msu.edu/adami.html

Adami expounded upon these ideas by defining life as "a phase of matter in which the dynamics of information processing overcome those of ordinary matter and energy." I am paraphrasing, and possibly butchering it a little... can't find the original quote, but that's the basic idea. Life is a phase of matter -- one whose behavior is dominated by universal Turing-complete computation. You might call living matter "Turium" or something.

The fact that this whole line of reasoning hasn't been taken up more broadly is IMHO some kind of failure of the academic and scientific system.

Made a jsfiddle where you can see the ant move and die (crash) after building the highway for a while.

http://jsfiddle.net/zuu7bd2e/

Thanks! It was hard to get the point of how cool this is from the animation/description in the Wikipedia article.

Langton's ant colony eBook cover generator: each word in the eBook title seeds the initial parameters (location, direction) of an ant and the simulation is run for a large, constant number of steps, producing an interesting and reproducible generated cover.

Top Ten Signs You've Been Reading Too Much Hacker News

Interesting "art" can be generated by running the algorithm against a random initial map.

Edit: adjusting the density of the initial map seems to result in different patterns appearing in the output

http://jsfiddle.net/2yxyno52/4/

How can it be Turing-complete with only two colors and one state? I thought you needed at least two colors and three states to be Turing-complete. https://en.wikipedia.org/wiki/Wolfram%27s_2-state_3-symbol_T...

2d versus 1d Turing machine apparently. Here is the paper claiming universality: http://www.dim.uchile.cl/~anmoreir/oficial/langton_dam.pdf

I believe that it is the distinction between playing on a 1-dimensional or 2-dimensional grid.

EDIT: Also, as the article that you link points out, while all definitions of Turing machines agree on their power as a class, minimality results like this are quite sensitive to the details of the definition. (In particular, the article seems to hedge its bets by saying that that machine may be the smallest universal machine.)

Hmm, I'm not convinced that there should be a difference between 1d grids and 2d grids, since 2d grids are countable and can be represented as 1d grids (just spiral out from any tile on the 2d grid and you'll get a 1d grid. The rules for the ant would have to take this into account, but I don't see any reason why they shouldn't be able to).

That's mostly just my intuition though; it is very possible that I am wrong.

Sure, but you'd need more internal states to deal with the overhead of that simulation. On a 2D grid "up" is a primitive op; in your 1D spiral, "up" is some complicated function that probably needs unbounded memory of its own to compute (i.e., an explicit pair of grid coordinates, stored as an integer).

(There's a textbook example where you can add auxiliary memory to a (1D) Turing machine, called a "multi-tape" Turing machine. And you can reduce that to a single-tape Turing machine, but at the cost of blowing up the number of states in the finite automaton).

In a sense it really has four "states", since the ant remembers which direction it had previously moved. That's implicit in the clockwise/counterclockwise rule.

I would say there are 10 states. Any square can be empty or have an ant, and if the ant is there its direction is known. That gives 5 states of ant presence and direction, and since a square has 2 color states, that's 10 total states.

I'm counting "states" as the term's used in (tape) Turing machines, where you distinguish the internal states of the finite automaton (tape head) from the memory states of the unbounded tape (symbols or colors). So this ant would be analogous to a 4-state, 2-color Turing machine. The ant has four possible states; each cell of the grid has two.

What about when the ant reaches the edge of his grid? (this part I couldn't find an explanation for on the wiki)

The grid has no edges, it's infinite in all directions!

I see. It would be interesting to see this on a finite universe - a curved universe where opposite edges join

Oh yeah, that's a good point!

> All finite initial configurations tested eventually converge to the same repetitive pattern, suggesting that the "highway" is an attractor of Langton's ant, but no one has been able to prove that this is true for all such initial configurations

If Langton's ant is capable of universal computation, wouldn't non-halting programs be a counterexample to this convergence?

I'd have to agree. From my understanding of that sentence, all programs would "halt" eventually because they all end up doing the highway. So unless your computation somehow ended with the highway then you couldn't run infinitely long computations.

Langton's ant has no halting state. It can't halt. But you're right, that description doesn't leave any room for programs that run in a specific loop forever instead of making a highway. So that doesn't seem to be Turing-complete.

One of the first programs I wrote was a simulator of Turmites in Visual Basic. There is an endless variety of configurations producing crazy abstract "art". I didn't know about Langton's ant back then, but I definitely noticed those "highways" that occured quite often from random experimentation.

I'm looking for an algorithm that is as simple as Langton's ant but with X amount of ants that all compete in some way. Any pointers?

The Wikipedia article contains several variations, including that one. Check the external links for videos and programs.

Interestingly, multiple ants don't need conflict resolution, as two ants sharing the same cell will want to leave it at the same state.

https://en.wikipedia.org/wiki/Langton's_ant

Thanks. I am actually looking for one with conflicting resolutions though and can't seem to find that in the links. Basically I want to make a game where you put out ants that compete somehow. Either the one with most converted tiles wins or if they can kill each other. It doesn't have to be Langton's ant specifically, just an algorithm that is as easy to understand.

I haven't thought this through, but I think a simple extension which would generate conflict would be to incorporate the ant's last movement in determining the end state of the cell it's currently on. That way, two ants arriving in the same cell from different directions would want different results.

Order in chaos. Love it. I have few different patterns for ants on my homepage (implemented in ugly ClojueScript).

How about 3D Langton's (or Langtonish) ants? Or more dimensions? :)

Need at least one new rule and a new color.

You could still do it with two colors if it remembered the previous color as well, and if you didn't allow "keep going straight" or "move back in the direction you came from" as options.

(Imagine the ant as a space-ship, and every time it turns, it also changes its pitch/yaw, so the directions are based on its direction of travel.)

    Current | Previous | Action
    white   | white    | go starboard
    white   | black    | go port
    black   | white    | go down
    black   | black    | go up

Reply on news.ycombinator.com