Page 1 of 1

Switch engine puffer search script

Posted: January 29th, 2012, 3:30 pm
by Jason Summers
Here's a Golly script I wrote to search for new switch engine puffers; i.e., anything other than the p288 "block maker" or the p384 "glider producer".
seo-search.zip

This is a long shot at best -- I've mostly given up on it, but I think there's still a chance of success.

It requires Perl scripting support. Read the comments at the beginning of the script for more information.

Re: Switch engine puffer search script

Posted: January 29th, 2012, 3:50 pm
by Wojowu
Is there a chance for Python version?

Re: Switch engine puffer search script

Posted: January 29th, 2012, 7:53 pm
by 137ben
It's nice, but what might be more useful is a similar search for other rules. It is only feasible to find small puffers, and life has been well searched.

Re: Switch engine puffer search script

Posted: January 29th, 2012, 10:49 pm
by Jason Summers
Wojowu wrote:Is there a chance for Python version?
It ought to take about half an hour to translate the script to Python. Maybe someone will volunteer? I could probably do it, but I'm a lot more comfortable with Perl than Python.

Re: Switch engine puffer search script

Posted: January 29th, 2012, 10:55 pm
by Jason Summers
137ben wrote:It's nice, but what might be more useful is a similar search for other rules. It is only feasible to find small puffers, and life has been well searched.
I've used random-perturbation scripts like this to find a number of new stable "orbits" into which a puffer's output can fall. The puffer doesn't have to be small, though of course the search will be faster if it is.

I don't think I tried any other rules, but there's no reason it wouldn't work. The real difficulty is that, for each puffer engine, you pretty much have to do custom programming to filter out the known orbits (including any sufficiently-common new ones you may discover). If all the orbits have convenient periods, you can mostly identify them by number-of-cells-added-per-period. But when you need something more sophisticated than that, it can get a lot more difficult.

Re: Switch engine puffer search script

Posted: January 30th, 2012, 9:54 pm
by 137ben
The real difficulty is that, for each puffer engine, you pretty much have to do custom programming to filter out the known orbits (including any sufficiently-common new ones you may discover). If all the orbits have convenient periods, you can mostly identify them by number-of-cells-added-per-period. But when you need something more sophisticated than that, it can get a lot more difficult.
It seems it would be best suited to searching for puffers in rules for which infinite growth is not currently known. This avoids having to "filter out" known puffers.

Re: Switch engine puffer search script

Posted: February 1st, 2012, 8:51 pm
by Jason Summers
To clarify, this is not a way to find completely new puffers. You have to start with a known puffer or near-puffer, and it will try to find different ways that its smoke can stabilize.