How about faster unidimensional spaceship

Post by **apg** » January 12th, 2026, 8:25 am

Hippo.69 wrote: January 4th, 2026, 5:49 pm So here it is:
795 435 669x1 2c/28 409 785 756:

Very impressive! I've created a wiki page for the new spaceship, https://conwaylife.com/wiki/Unidimensional_spaceship_2

It's such an enormous improvement over the previous version (almost 5 times smaller by period, population, and bounding box) that you can now run an entire period in Golly in 15 minutes at a step size of 8^7.

Hippo.69 · Post by **Hippo.69** » January 12th, 2026, 9:52 am

apg wrote: January 12th, 2026, 8:25 am Very impressive! I've created a wiki page for the new spaceship, https://conwaylife.com/wiki/Unidimensional_spaceship_2

It's such an enormous improvement over the previous version (almost 5 times smaller by period, population, and bounding box) that you can now run an entire period in Golly in 15 minutes at a step size of 8^7.

Thanks, I am still working on it, the fuse arm optimization is almost done, I hope that after realigning and small change in cleanup, it would work as the version 2 without further changes, but due to changed distances, the semistate configuration after ecca1 stop could differ ... .
But the changes would be negligable compared to transition from version 1 to version 2.

I6_I6 · Post by **I6_I6** » January 12th, 2026, 10:08 am

Hippo.69 wrote: January 4th, 2026, 5:49 pm I got at epicentre this at generation 28 409 785 856: (otherwise unidimensional)
Code: Select all
x = 46, y = 21, rule = B3/S23
4$39bobo$39b2o$40bo4$35b2o$2b3o2b3ob3ob3ob3ob3o2b3o3bo2bo$35b2o4$40bo$
39b2o$39bobo!

Huh? I don't understand. So is it unidimensional?
Also, how do I add pattern files to the infoboxes of LifeWiki articles?

Hippo.69 · Post by **Hippo.69** » January 12th, 2026, 10:12 am

I6_I6 wrote: January 12th, 2026, 10:08 am Huh? I don't understand. So is it unidimensional?
Also, how do I add pattern files to the infoboxes of LifeWiki articles?

This was last unsuccessfull attemt to be corrected ... and I have reedit the post to present working version ... actually 100 ticks faster (the phase of the last cordership was changed and the phase of the ott as well).

Hippo.69 · Post by **Hippo.69** » January 12th, 2026, 9:12 pm

Oops, seems I have not checked in correct final version of rmac.py of unidimensional ship 2 to git ... and I have overwritten it with older version from git before making changes in cleanup ... . I should reinvent how to stop binary arm gun during cleanup (as it cannot be stoped during rebuild) ...

Reinvented ... checked in, pushed.

Here is current frobenius (798415x1):

frobenius.mc: Improved fuse arm optimizing rebuild cost, bigger repertoire of fuses; (144.03 KiB) Downloaded 33 times

Or rather this (798357x1):

frobenius.mc: Optimalized fuse start; (143.63 KiB) Downloaded 33 times

Or rather this (796355x1):

frobenius.mc: removing 1+1 gliders from 2 cleanups; (142.61 KiB) Downloaded 34 times

I6_I6 · Post by **I6_I6** » January 14th, 2026, 1:19 pm

Hippo.69 wrote: January 4th, 2026, 5:49 pm So here it is:
795 435 669x1 2c/28 409 785 756:
Maximal sizes about 2 360 076 000x2 359 802 000

Glider synthesis:

1Dship2_synth.mc: (3.57 MiB) Downloaded 44 times

Like when I synthesized the first unidimensional spaceship, I don't know how much the tape costs. I only know that the rest of it costs 207936 gliders.

EDIT:

Hippo.69 wrote: January 12th, 2026, 9:12 pm Oops, seems I have not checked in correct final version of rmac.py of unidimensional ship 2 to git ... and I have overwritten it with older version from git before making changes in cleanup ... . I should reinvent how to stop binary arm gun during cleanup (as it cannot be stoped during rebuild) ...

Reinvented ... checked in, pushed.

Here is current frobenius (798415x1):
frobenius.mc

Or rather this (798357x1):
frobenius.mc

Or rather this (796355x1):
frobenius.mc

So, are those improved versions of the fuse? Doesn't that mean the tape needs to be changed, too?

Hippo.69 · Post by **Hippo.69** » January 14th, 2026, 3:05 pm

I6_I6 wrote: January 14th, 2026, 1:19 pm Glider synthesis:
1Dship2_synth.mc

Like when I synthesized the first unidimensional spaceship, I don't know how much the tape costs. I only know that the rest of it costs 207936 gliders.

Oh nice, seems two directional salvo can build the ship in cost 2 gliders per blinker (using 4 directions is confusing). So number of gliders corresponding to 2/3 population (except the start).

I6_I6 wrote: January 14th, 2026, 1:19 pm
Hippo.69 wrote: January 12th, 2026, 9:12 pm Here is current frobenius (798415x1):
frobenius.mc

Or rather this (798357x1):
frobenius.mc

Or rather this (796355x1):
frobenius.mc
So, those are improved versions of the fuse?

Yep, I am locating the offset for ECCA2 inclusion ... and if I would be lucky the new version would be ready (just ecca2 should destroy one more boat close to the arm). (Version 2 had frobenius/fuse size 827733x1, but more importantly uses expensive blinker distance 4 or 5 much more often then the current one).

Hmm, end of ecca1 program (beehive, ship, transition to ecca2) should be updated as well ...

I6_I6 · Post by **I6_I6** » January 14th, 2026, 3:42 pm

Hippo.69 wrote: January 14th, 2026, 3:05 pm Oh nice, seems two directional salvo can build the ship in cost 2 gliders per blinker (using 4 directions is confusing).

The reason I used 4 directions is so that I could fit in the fuse ignition between the fuse and the tape without crossing glider lane issues. By the way, the synthesis synthesizes the phase that is 549,148 generations after the 1D phase of the spaceship (when preliminary fuse finishes).

Hippo.69 wrote: January 14th, 2026, 3:05 pm So number of gliders corresponding to 2/3 population (except the start).

You're right; that means the blinkers on their own cost 4,096,896 gliders, so the full thing costs 4,096,896+2 (a TL)+2 (fuse ignition)+4 (kickback for 2 gliders which will hit the 2 outward LWSSs)=4,096,904 gliders

EDIT:
Similarly, unidimensional spaceship 1 costs 20,370,462 gliders because it was synthesized exactly the same way. That's almost 5 times more expensive!

ElijahKen · Post by **ElijahKen** » January 15th, 2026, 11:24 am

I6_I6 wrote: January 14th, 2026, 1:19 pm
Hippo.69 wrote: January 4th, 2026, 5:49 pm So here it is:
795 435 669x1 2c/28 409 785 756:
Maximal sizes about 2 360 076 000x2 359 802 000
Glider synthesis:
1Dship2_synth.mc

Like when I synthesized the first unidimensional spaceship, I don't know how much the tape costs. I only know that the rest of it costs 207936 gliders.

It costs 4096904 gliders (Edit: total).
EDIT 2:
I hadn't noticed the previous post which came to the same conclusion, sorry.
I came to this conclusion by downloading the synthesis file, loading it into lifelib and counting the population, then dividing by 5.
It should be possible to greatly reduce this by finding a spaceship convoy that can turn 1 glider into a blinker, or similar.

Hippo.69 · Post by **Hippo.69** » January 15th, 2026, 8:32 pm

Hmm, I am still strugling with the end of the fuse reconstruction (the wrong work with git

...), and the ECCA2 is wrongly positioned ... it would take some time to finish it ...

OK, end of fuse reconstruction corrected (and comitted). So shifting ecca2 is the next step and than we see ... how ecca1 -> ott changes.
And ECCA1 would be definitely stopped in another state so the conversion to ott should be recomputed.

Even the GPSE90 stop needs another synchronization (gliders should be delayed 158 ticks).

So ECCA2 should be shifted 8fd further from GPSE90's, salvo destruction for ECCA1 recomputed, GPSE90 stopping seed start delayed by 158 turns, adding a glider to Near cleanup ... to destroy an extra boat ... and it could be all.

... I still have not checked the other direction in ECCA2 positioning, but there is no reason to have a difference ... (fuse arm changed timing, so positions in one direction as well, but why the other?).

So ecca2 repositioned, boat destroyed, stop puffers delay adjusted and ecca1 destruction is computed (still computed in golly what takes time) ...

Hmm, GPSE90 stopping requires more attention ... individual GPSE's are not started at original delays so all "4" puffers should be resynchronized (3 GPSE'90 and pair of backfiring "corderships").
How much should be 2nd GPSE stopper more dealyed relative to the first stopper? ... Good question ... 42 ticks.
For the 3rd GPSE stopper the difference to first is 66.
What is the difference for the 'corderships'? It could stay as it is (there is small tollerance).

OK ECCA1 destruction finally computed ... "integration testing" starts.
As expected the size dropped only a little to 776437671x1.

Hippo.69 · Post by **Hippo.69** » January 19th, 2026, 4:22 pm

I6_I6 wrote: January 14th, 2026, 3:42 pm
The reason I used 4 directions is so that I could fit in the fuse ignition between the fuse and the tape without crossing glider lane issues. By the way, the synthesis synthesizes the phase that is 549,148 generations after the 1D phase of the spaceship (when preliminary …

You probably can use salvo from east to construct the sleeping pattern and ignite it as well.

Hippo.69 · Post by **Hippo.69** » January 30th, 2026, 5:47 pm

So the 3rd version

example.mc: (2.24 MiB) Downloaded 67 times

776 525 415x1 2c/27 735 693 820.

Better fuse arm, but this didnot lead to a significant improvement (as expected).

There could be chance for making a ship significantly faster for the cost to be a little bit longer … stopping gpse90 by ecca1 rather to waiting for ecca2.

The program length is still significantly smaller then the maximal ship length … stopping the GPSE90’s close to the tape length would help.

Now more then 1/3 of the ship length is building the ECCA1 so improving binary recipe could become visible, but stopping the tapereading earlier has higher priority. Hmm, let us hope optimal timing for the stop does not depend on binary arm program ... but if there would be different ecca1, the ecca1 destruction would differ what affects ecca2 program ... but is currently only about ecca2 650 gliders so about 5 000 bits ... 1 000 000 pixels.

Sending synchronized stopping salvo near the end of building ecca2 could work.

OK 4th step would be earlier stop, 5th step binary program replacement.
But simulatable 0e0p metacell has higher priority.

So this should be included in cleanup phase: