Thanks for checking! I didn't really expect anything workable would show up in the 3G collisions -- even if the ash signature had happened to show up, quite likely the reaction wouldn't have been compatible with the other half of the loafer recipe.2718281828 wrote:If I got it correctly the relevant part of the 5 glider reaction isI looked for the resulting ash (including symmetries) in my 3G collision files, but the result is negative.
Code: Select all
x = 9, y = 4, rule = B3/S23 b2o4b2o$o2b2ob3o$bobo$2bo!
I have somewhat higher hopes for a workable solution with four gliders, since the search space is several orders of magnitude bigger. It seems like a good goal will be to produce a set of gencols input files, with a verified list of all the relevant 3-glider collisions of each type, say 3G-NW-NW-NE.col and 3G-NW-NE-SW.col.
Maybe sorted in order of the population of the final output, or something along those lines? That would be a good fraction of the work needed to set up a "hashseq" database and search utility. To limit the size of the database, the 3G collision files could be truncated at some reasonable point, leaving out all the almost certainly useless huge ugly explosions, and then pick a reasonable range of locations for the fourth glider.
-- Thinking about it again, it's probably worth keeping 72 separate sets of 3G collisions, 71 for the different 2G collisions and one more for all the simultaneous ones, and split each of the 72 sets up into categories along the lines of NW-NW-NE and NW-NE-SW. That allows 4G searches to be customized for each group, over time, and a lot of those could be exhaustive searches.
(Apologies -- I've wandered off into hand-waving about vaporware again.)