This tutorial describes a few stages of the process of building a 1G seed -- an engineered constellation of small glider-constructible still lifes, designed so that when a single glider collides with it, a complex target object (spaceship, rake, wickstretcher, etc.) is eventually produced.
The usual underlying mechanism is either a standard glider synthesis recipe, or a modification of a standard recipe that includes additional bait objects to reduce the number of synchronized gliders needed. To turn a single incoming glider into any number of synchronized gliders, we can use one-timeturners and splitters.
For several years standard resource for one-time turners and splitters has been Michael Simkin's collection of constellations fitting inside 10×10, on the Splitters with Common SL thread. However, these constellations included less common still lifes such as shipties and mangoes, and also allowed pseudo-objects, which are somewhat more expensive to construct.
A more recent toolkit of 90-degree turners can be found later in the same thread. The classification of 90-degree turners is complete and correct, but as of September 2020 the 0-degree and 180-degree turner categories are a work in progress -- even "CP" and "CC" labels are inverted in some cases, and there should be separate categories for each output lane. The 2-glider and 3-glider splitters are currently uncategorized; each should be labeled according to the two or three categories of turner that its output gliders fall into.
Overview
What follows is a walkthrough of the first part of the construction process for finishing a c/4 diagonal wickstretcher seed for use in self-constructing circuitry. The walkthrough is based on this forum posting, which in turn was based on some notes from Discord.
a reaction that provides the necessary spark to start the wickstretching reaction behind the already-constructed crab
a "guard glider" that, once the wick has been set to burn at 2c/3, changes the way the burning fuse interacts with the crab when it catches up with it, destroying the crab and producing a minimum of ash and no output gliders.
For this project we'll use the wickstarting reaction twice, once on each side. We will also add a different guard glider on each side. The guard glider on the northwest side allows the wick to be burned without destroying the crab spaceship or the other wick, leaving behind a target object (a block) for a following slow salvo to make use of. The guard glider on the southeast is in a good location to allow the fuse to stop the crab spaceship, destroying it in a controlled way without releasing gliders, and again leaving a target behind.
x=68, y = 85, rule = LifeHistory
58.3A$60.A$59.A$56.2A$56.A2$54.A3.2A$53.A.A2.2A5.3A$52.A.A3.A.A6.A$
51.A.A6.2A4.A$37.3A10.A.A6.A.4A$39.A9.A.A10.2A$38.A9.A.A8.A$47.A.A8.A
.A$46.A.A8.A.A$45.A.A8.A.A$44.A.A8.A.A$43.A.A8.A.A$42.A.A8.A.A$41.A.A
8.A.A$40.A.A8.A.A$39.A.A8.A.A$38.A.A8.A.A3.3A$37.A.A8.A.A6.A$36.A.A8.
A.A6.A$35.A.A8.A.A$34.A.A8.A.A$33.A.A8.A.A$32.A.A8.A.A$31.A.A8.A.A$
31.2A8.A.A$40.A.A$39.A.A$39.2A2$16.C$15.C.C$16.2C5$5.3C$7.C3.2C$6.C3.
C2.C$11.2C2$32.2E$32.E.E$33.E2$25.E$24.E.E$24.E.E$25.E28$.2E$E.E$2.E!
[[ THEME LifeHistory GRID TRACK 1/4 -1/4 STOP 930 X 0 Y -15 Z 3 WIDTH 600 HEIGHT 600 THUMBNAIL THUMBSIZE 2 ]]
The two target objects can be separated by as far as we want by delaying the second (yellow) glider relative to the first (white) glider.
Outline
Starting from a non-working previous attempt at a seed, here are the steps that we'll follow:
remove the incompatible splitter (two yellow eaters) and put in something that won't get in the way as much;
connect up the top glider with a very fast turner, block off the other glider, get the wickstretcher activation working;
put a NW-traveling glider in the right place to activate the guard glider correctly (right timing and everything);
put in either a fast CP0 or fast CC0 turner to turn the lower glider from the splitter onto that same lane, and see whether it's off by 0 1 2 3 4 5 6 or 7 mod 8;
put in a fast CXn instead of the CX0 -- a CP1, for example, will be one tick slower than a CP0;
try it out, then count how many times 8 ticks it's off by, cross your fingers, and go and see if you have any turners in the right bucket.
If at the last step it happens that there aren't any turners available with exactly the right timing, one thing that could be tried is to shift the splitter over by one cell, changing the colors of the output gliders. Then retry the same steps. This will mean using all different 90-degree turners -- CP turners instead of CC ones, and vice versa -- so you get a second chance to have the timing work out.
The yellow and white OTTs each work fine in isolation, but they can't be combined -- the yellow splitter made out of two fishhook eaters is much too slow and takes up too much space.
This means that we're going to need a different way to produce the two yellow gliders below from one NE-traveling glider:
Instead of the two-eater glider splitter, et's try using this eater-and-longboat splitter -- it produces its gliders much quicker and uses much less space (smaller reaction envelope):
Where it stops at T=84, you can see that the glider is getting there 10 cells too soon -- 40 ticks. So move the NE-traveling yellow glider back by 10 cells, and delete the glider that it's replacing. Now we've successfully synchronized one of the two gliders; that signal is now coming from the southwest via a splitter, replacing the temporary reference glider that came from the southeast:
Already we've reached the point where we just need to find exactly the right 90-degree one-time turner to replace the remaining NW-traveling glider, using the glider from the eaternlongboat splitter that's been blocked off (pun intended) by the red-and-yellow preblock.
You can see that the turn is a color-preserving turn, because there's just one cell on the inside of the turn (above the preblock). Here's a possible mnemonic: "CP" has one "C" in it, and there's one cell on the inside corner of a color-preserving turn; "CC" has two "C"s in it, and there are two adjacent cells on the inside of of a color-changing turn.
With a lot of cleverness we could figure out which category of turner would turn the NE-traveling glider into the NW-traveling glider. But it's easier to just do it by experiment. Simply put in a CP0 turner and see what happens, then replace it as necessary with a turner in a different category.
This is the fastest turner in the CP0 category, a CP0col9:
Run it and you'll see that a CP0col9 is exactly 30 ticks faster than what we need -- the glider gets to its destination too soon.
That means a CP6col9 would be exactly 24 ticks faster than what we need, so the turner we want is going to be in the CP6 bucket. Furthermore, every column "costs" four ticks (because gliders move at c/4) -- so the turner we really want is one that's rated CP6col15, six columns, 6*4 = 24 ticks, slower than col11.
However, when we try a CP2col15, it's still eight ticks too fast. This is a limitation of the current categorization system. A CP6 glider has traveled 6 ticks from the CP0 location, so it's already two columns over from its CP0 equivalent. Maybe the classification script will be adjusted in the future to account for this. But in the meantime, what we really want must be a CP6col17 --
The whole structure can then be tightened up a little bit. One-time turners modifying c/4 spaceships are very forgiving -- for example, you could move all the yellow objects except for the yellow activation glider northeast as far as you want, and the mechanism will still work perfectly well.
So a final candidate 4G (four-glider) seed would be
x=96, y = 92, rule = LifeHistory
82.3D$84.D$83.D$80.2D$80.D2$82.2D$82.2D5.3D$82.D.D6.D$84.2D4.D$83.D.
4D$86.2D$83.D$82.D.D$81.D.D$80.D.D$79.D.D$78.D.D$77.D.D$76.D.D$37.2A
36.D.D$37.2A35.D.D$73.D.D3.3D$72.D.D6.D$71.D.D6.D$70.D.D$30.A38.D.D$
28.3A2.2A33.D.D$27.A5.A.A31.D.D8.2E$27.2A6.A30.D.D10.E$35.2A28.D.D8.
3E$65.2D9.E$26.A17.2A$25.A.A9.2A5.2A$24.A.A11.A10.2A26.2E$24.2A12.A.A
3.2A2.A.A25.E2.E$39.2A2.A2.A2.A12.2D13.E.E$43.A.A16.2D7.2E5.E$44.A26.
E$66.2D.E.E$48.A17.2D.2E$47.A.A5.2A$43.A3.A2.A3.A.A29.2E6.E$37.2A4.A
4.A.A4.A20.2E8.E.E4.E.E$37.A.A3.A5.A21.E4.E.E8.E.E4.2E$39.A14.A15.E.E
4.2E9.E$39.2A12.A.A.2A12.2E$46.3A3.A2.A.2A$53.2A22.E$76.E.E5.2E$75.E
2.E5.2E$54.2A20.2E$2A52.A$.2A52.3A11.2A$A56.A11.2A$60.A$60.3A$63.A$
62.2A14.E$71.D6.3E$70.D.D.E6.E$63.2A6.2DE.E4.2E$63.A10.E.E$61.A.A11.
2E$57.A3.2A$56.A.A$55.A.A$55.2A9$54.2E$53.E.E$55.E$6.A$6.2A$5.A.A8$
37.2A$38.2A$37.A!
[[ THEME LifeHistory GRID WIDTH 600 HEIGHT 600 THUMBSIZE 2 THUMBLAUNCH X 23 Y 1 Z 4 ]]
We can repeat the same kind of analysis to add more splitters and synchronize their output gliders, creating the four synchronized gliders from one single input glider. Here's a sample completion of the project to make a 1G seed for the double wickstretcher: