how about a sir robin synthesis?

For discussion of specific patterns or specific families of patterns, both newly-discovered and well-known.
Post Reply
Posts: 219
Joined: January 24th, 2018, 4:36 pm

how about a sir robin synthesis?

Post by googleplex » March 7th, 2018, 11:07 am

Welp, you knew it was coming! can we synthesize it?
Last edited by googleplex on March 8th, 2018, 9:29 am, edited 1 time in total.
Look at me! I make patterns in golly and go on the forums! I wanna be Famous!

User avatar
Posts: 744
Joined: March 29th, 2009, 4:58 am

Re: Synthesizing the elementary knightship

Post by Macbi » March 7th, 2018, 12:25 pm

Maybe we should try for an eater before we look for a gun.

User avatar
Posts: 6873
Joined: May 17th, 2009, 11:00 pm
Location: Madison, WI

Re: Synthesizing the elementary knightship

Post by dvgrn » March 7th, 2018, 12:35 pm

Macbi wrote:Maybe we should try for an eater before we look for a gun.
An eater won't be a problem -- I could build one in a few hours. A compact or elegant eater, on the other hand, might be a bit tricky... Luckily we probably don't need to worry about an eater with the fastest possible recovery time, until and unless a knightship gun actually does show up.
googleplex wrote:Welp, you knew it was coming! can we synthesize it?
Not today!

(I have a mixed record on Life-related prognostication, but I'm pretty confident on this one.)

There are many thousands of known spaceships that we don't know how to synthesize, that would be orders of magnitude easier to synthesize than this huge fast-moving high-period object. You almost might as well try to synthesize a spaghetti monster.

That's not to say that someday we might not have a backtracking search algorithm that generates predecessors, and recursively chooses predecessors that are more amenable to being constructed by gliders -- maybe they contain large simple stable regions, or maybe the search will prefer predecessors that effectively break the construction problem into several widely-separated smaller islands.

The problem is to define a metric -- an algorithm that can somehow give each predecessor an accurate score, to tell us reliably that one predecessor is somehow "closer" than another to being a glider synthesis. People have done a lot of hand-waving about what might work as a metric, but so far it's all vaporware.

Once we have actual code that assigns a helpful score to an arbitrary pattern, the rest of the search becomes relatively easy -- assuming that it will usually be possible to find predecessors for a given pattern that have a lower score than the pattern itself. That might be a big assumption.

Without a reliable metric, the glider synthesis problem for this particular knightship is -- I would guess -- billions of times harder (at least) than the problem of finding the knightship in the first place... and that only happened after Tom and Adam ran multiple-core searches over a period of several weeks.

User avatar
Posts: 1497
Joined: January 12th, 2018, 9:19 pm

Re: Synthesizing the elementary knightship

Post by 77topaz » March 7th, 2018, 3:32 pm

I think, because this spaceship is very large and contains a lot of space dust, it would be very difficult to synthesise.

Post Reply