codeholic wrote:I had the following design in mind for quite a while, but I've drawn it only now. I wonder how feasible it is. It seems hardly possible to find its solution without computer assistance, but unfortunately I'm not so deep into 31c/240 arithmetics to make a program on my own.
I think you're right that a custom search program would be really handy, either for completing your design, or for finding useful ways to put kinks in the block trails as Extrementhusiast suggested.
The easiest way to bend a block trail is with a (2,1) block pull. Simple Herschel-pair climbers work on [54..59]+31N, with some weird options for [60,61]+31N... and for some of these distances, they work even if the block trails aren't exactly aligned, so it might be possible to (2,1) block-pull them back and forth, and still have them work -- i.e., it would be possible to send Herschel pairs up the block trails both before and after the adjustment:
Code: Select all
#C 54-spaced block trails turning rather awkwardly into 56-spaced trails
x = 951, y = 482, rule = LifeHistory
445.2C52.2C$445.2C52.2C30$445.2C52.2C$445.2C52.2C30$445.2C52.2C$445.
2C52.2C30$445.2C52.2C$445.2C52.2C30$445.2C52.2C$445.2C52.2C30$445.2C
52.2C$445.2C52.2C18$449.2A$447.2A.A$447.2A$449.2A$451.2A38.3A$492.A$
450.A.A39.3A$451.A5$445.2D52.2D$445.2D52.2D2$501.2A$453.2A36.2A8.2A$
453.2A36.2A5$503.A$502.A.A$501.A3.A$501.A3.A$501.A$502.3A7$445.2A52.
2A$445.2A52.2A8$445.2D52.2D$445.2D52.2D3$453.2A36.2A$453.2A36.2A16$
420.2A$421.2A131.A$420.A131.A.A$444.2A54.2A51.2A$444.2A54.2A$528.3A$
386.A.A139.A$386.2A141.A$387.A$383.2A$384.2A$383.A3$565.3A$565.A$566.
A10$438.A.A$438.2A$439.A$360.2A$361.2A148.A$360.A151.A$510.3A2$444.2A
54.2A86.3A$444.2A54.2A86.A$589.A2$323.2A$324.2A$323.A3$625.3A$625.A$
626.A13$300.2A$301.2A$300.A3$648.3A$648.A$444.2A54.2A147.A$444.2A54.
2A$263.2A$264.2A$263.A3$685.3A$685.A$686.A13$240.2A$241.2A$240.A3$
708.3A$708.A$709.A2$203.2A239.2A54.2A$204.2A238.2A54.2A$203.A3$745.3A
$745.A$746.A13$180.2A$181.2A$180.A3$768.3A$768.A$769.A2$143.2A$144.2A
$143.A300.2A54.2A$444.2A54.2A2$805.3A$805.A$806.A13$120.2A$121.2A$
120.A3$828.3A$828.A$829.A2$83.2A$84.2A$83.A2$444.2A54.2A$444.2A54.2A
363.3A$865.A$866.A13$60.2A$61.2A$60.A3$888.3A$888.A$889.A2$23.2A$24.
2A$23.A3$925.3A$444.2A54.2A423.A$444.2A54.2A424.A13$2A$.2A$A3$448.2A
498.3A$446.2A.A498.A$446.2A501.A$448.2A$450.2A40.3A$493.A$449.A.A41.
3A$450.A9$452.2A38.2A$452.2A38.2A17$444.2A54.2A$444.2A54.2A!
This might end up being the basis for more new rakes, come to think of it (arrrgh). But it doesn't seem to work as well when you shoot down the "main block" and leave the "spare block"; the offset is too big, so I haven't come up with any before-and-after spacings that do anything particularly useful yet. Maybe there's a way to get a pair of trails to drift gradually closer to another pair of trails, though, using (2,1) block pulls -- I'll investigate that a little more. A back-and-forth (2,1) or (1,2) drift would reset the phase by +/-2 or +/-4, which could also be very useful -- it takes a lot of rephasers to do that... Then again, it will take a dozen precisely-aimed gliders to do that for a full six-trail track, so probably the rephasers will be more efficient!
There are other trail separations that work
only if the trails are not precisely aligned, but usually those fail to clean up one spare block or the other on one of the trails, and produce an extra output glider instead. So things get complicated very quickly. A brute-force search utility might do about as good a job at digging up new combinations as my slow and imperfect half-theoretical half-experimental approach...!
No doubt there are more complex arrangements of block trails where three or four or five trails with various offsets cooperate to clean up each other's spare blocks, just not in simple pairs. There's probably even a four- or five-trail setup that can support forward and backward rakes somehow -- just nobody has found it yet.
codeholic wrote:Once you have these six trails, you can build the next, narrower set to make things faster.
Somehow I suspect that the shortest possible 31c/240 spaceship will build just six block trails, and use those to support whatever rakes can be supported. Once you have six trails, there's not much point in wasting spaceship length by building more -- unless they're very very widely spaced, I suppose. But it seems as if we might just as well build reasonably closely spaced trails to start out, and be done with it.
As usual, I'll be very happy to be proved wrong about this, but probably won't spend much time or effort trying to prove myself wrong...!
The 116:110:90 trail spacing, which supports most of the currently known rakes, was a fairly arbitrary choice. There are definitely good arguments for other spacings, including arrangements where not all of the trails are aligned. Many of the known rakes could no doubt be adapted to work on other spacings.
In particular, a
nice simple standard 116-spacing still has a lot going for it -- you can run climbers up any pair of adjacent trails, add as many more trails as you want with the same tools, and so on. I haven't dared to go back with our current collected knowledge and seriously investigate how much easier the 116x5 spacing would be -- don't really want to rebuild piles of rakes for the new spacing.