Finding patterns in GOL

[SystemsLock]

Ok, so hey everyone. I am sorta new to GOL. I know a fair bit about it (read a Wikipedia page on it) and have been fooling around for a few hours on Golly. So anyway I have seen many different patterns, especially with the more complicated one (like the UTM) how the heck do you build them? I mean, do you just experiment with existing patterns untill you get something new? Do you use basic knowledge about the game and build up? or are you using brute force computers will some complex algorithm? Honestly aside form guessing and checking I surely have no idea how anyone can create a pattern on purpose. Anyway, anyone who can help me please drop by a post. I am excited to do more then watch other peoples patterns flash by.

-System

[H. V. McIntosh]

..... Do you use basic knowledge about the game and build up? or are you using brute force computers with some complex algorithm? .....

Doubtless people are using all the methods you have suggested. When Life was new, people used to spell out their names using dotmatrix letters and then looking to see what happened. Polyominoes were one of Conway's favorites. But seriously, we are talking about forty years of experience and evolution here, during which time understanding and insight have greatly increased, computers and computer programs have become vastly more powerful, and the theory of automata in general has progressed.

For example, to understand the construction of a Turing Machine (UTM) would you want to be familiar with their purpose and how they operate? Otherwise how would you know what you were looking at? Anyway, there are lots of interesting little tasks to work on. Currently I am trying to decide how to follow up on a suggestion that quasars are extensible. For the moment that just consists of stacking pulsars together and seeing what happens. Using a program with limited screen space!

Although it is hard to recommend a single source explaining Life in all its detail, there are numerous sites on the Internet, even this very forum, and numerous links in its Wiki which can be followed up. Welcome to the party!

[calcyman]

Honestly aside form guessing and checking I surely have no idea how anyone can create a pattern on purpose.

The same way that you build an electronic circuit, or any other machine for that matter -- building it out of components. Any pattern functions identically no matter where it is positioned.

I myself have built numerous 'large-scale' devices, up to and including a universal computer. The majority of it was created with simple copy-and-paste operations in Golly. You can read about it on Heinrich Koenig and Dave Greene's website:

http://pentadecathlon.com/lifeNews/2009/08/post.html

[dvgrn]

..... Do you use basic knowledge about the game and build up? or are you using brute force computers with some complex algorithm? .....

A very large fraction of the more complicated patterns certainly had brute-force computer searches somewhere in their convoluted lineage. Just for example: the stable pseudo-Heisenburp pattern (Golly's Scripts/Python/heisenburp.py) has at its heart Dean Hickerson's 2c/3 wire, which I believe was found by Hickerson's 'drifter' search program. But the various signal tracks that connect to the ends of the 2c/3 wire were assembled with the help of Karel Suhajda's 'Hersrch' program, out of pieces that were either found by inspired experimentation (David Buckingham's original Herschel tracks from the mid-1990s) or by a brute-force search utility (Paul Callahan's 'ptbsearch'). There's a boojum reflector in there, too, which was found with a different program, Gabriel Nivasch's 'catalyst'. And Noam Elkies found the 2c/3 signal insertion reaction more or less by hand, no computerized searches at all.

Callahan's 'gencols' is often handy for figuring out glider recipes for things, or for coming up with new useful reactions involving sparks from oscillators rubbing against each other -- examples of this kind of thing can be found in Brice Due's metapixels and many other places. But of course once a discovery is made, it quickly becomes "basic knowledge" that other people can build on.

David Eppstein has a summary page describing Life search software used for various types of searches.

For this forum, I tried writing up a detailed description of how to use the Herschel-track search utility, Hersrch. Not sure if it came out comprehensible, though. I'd like to write sample walkthroughs for many of the other search utilities, too, over time -- gencols, ptbsearch, and some version of lifesrc, at least, and maybe 'drifter' after I understand it properly myself! But it's going to be a slow process at the rate I'm going so far.

[Extrementhusiast]

Many things, especially spaceship reactions, are actually found by accident!

[PM 2Ring]

I downloaded "catalyst" last week, but couldn't get it to compile under gcc on Linux. There was a missing header "profile.h". I commented that include statement out, and it compiled ok, but it wouldn't link. Any suggestions? (I can program in C, but know next to nothing about C++).

[Andrew]

I downloaded "catalyst" last week, but couldn't get it to compile under gcc on Linux. There was a missing header "profile.h". I commented that include statement out, and it compiled ok, but it wouldn't link. Any suggestions? (I can program in C, but know next to nothing about C++).

I also had to comment out the "#include <profiler.h>" line, but was then able to build it using these commands:

Code: Select all

g++ -Wall -c catalyst2.cpp
g++ -Wall catalyst.cpp -o catalyst catalyst2.o

[PM 2Ring]

Thanks, Andrew. I'll give that a try when I get home from work.

Edit.

Success! Now I just have to figure out how to use this thing. A few worked examples would be enormously useful...

[Guam]

Where can I download the Hickerson's program dr (drifter search)?

[Sokwe]

Where can I download the Hickerson's program dr (drifter search)?

I think you will need to contact Dean Hickerson. I will send you a private message with his contact information.

[dvgrn]

A few worked examples would be enormously useful...

I was reminded of this old 'catalyst' thread when it turned into an active topic again. Looking back through my old emails, I've found a version of 'catalyst' that I modified back in 2001 to look for stable reflectors, and then later fixed up with some extra documentation so that someone else might be able to understand it.

I called it 'catgl', since it's just 'catalyst' with some extra glider-detection and pattern-matching code cobbled on. It helped me find the boojum reflector, so maybe it's worth making available here. Source code, support files, and a Windows executable in a ZIP archive can be downloaded here.

There's a batch file in the archive that lets you run the program with a sample set of parameters. If anyone has difficulty getting it working (or downloading it, perhaps because of the archive's .BAT and .EXE files) just let me know...!

[dvgrn]

A few worked examples would be enormously useful...

I've found a version of 'catalyst' that I modified back in 2001 to look for stable reflectors... I called it 'catgl', since it's just 'catalyst' with some extra glider-detection and pattern-matching code cobbled on... ZIP archive can be downloaded here.

On another thread today I posted a sample search for a loafer-to-glider converter. It's not a very good example, because it doesn't start with any existing catalysts or make use of the various cell-type options. Sometime soon I'll add a single-catalyst sample search -- starting from three catalysts, find the fourth to suppress the traffic light and complete the loafer-to-glider converter. Something like that.