Macbi wrote:What's the easiest way in a Golly script to run a pattern to stabilization, detecting any escaping gliders?
Most often I've just borrowed the relevant pieces of the old pre-apgluxe apgsearch script, as toroidalet suggests.
Seems like I've rewritten Redstoneboi's method several times in different languages, including in C++ when Gabriel Nivasch's 'catalyst' got turned into 'catfind' to hunt for the boojum reflector. But hunting for live cells all the way around the edges of a pattern, and then hunting all over again a few ticks later, is a lot less efficient than just running the pattern for (big-HashLife-step) and using apgsearch's tricks to get rid of escaping gliders.
Macbi wrote:I actually had a clever idea for shrinking the size of the "seed" still life, down to only 110 cells!
Whatever complicated circuitry is needed, we can no doubt construct it using a (not slow) single channel salvo. So instead of actually using that circuitry as our seed, we can use a rectifier loop (population 110) large enough to store that recipe. Then we can begin our construction by loading our recipe into the loop, as slowly as we like. Then to get our elbow block and hand we explode one of the rectifiers by inserting a glider into the loop too close to another one.
So if let's say your definition of "slow" is "at least a billion ticks apart", then you can just make the rectifier loop a billion-tick loop. Then to get two gliders 106 ticks apart, you could send a glider at T=0 and the next at T=1,000,000,106. Nice!
Macbi wrote:I haven't actually done the hard work yet of trying each possible explosion to find one that places an elbow correctly without any escaping gliders, but I'm sure it won't be that hard. Considering that we can also have gliders in the loop take part in the explosion, there are thousands of possibilities. So I think it's very likely that one of them works.
Yeah, not worried about that either -- some combination is bound to work.
However, the rectifier has a repeat time of 106, whereas current single-channel recipes go down to 90-tick separation. We have simeks' assurance that higher repeat times are no problem -- probably to up above 200 ticks, let alone 150 or 106 -- but I'm not sure we have enough recipes available right now to prove universality.