Demonoid (diagonal Geminoid) completed!

For discussion of specific patterns or specific families of patterns, both newly-discovered and well-known.
User avatar
dvgrn
Moderator
Posts: 10612
Joined: May 17th, 2009, 11:00 pm
Location: Madison, WI
Contact:

Re: Demonoid (diagonal Geminoid) completed!

Post by dvgrn » October 21st, 2018, 9:26 am

chris_c wrote:One of these days I should try slmake and Gol_destroy on the 0hd Demonoid. From briefly looking in the past it should give close to a 50% improvement. The problem is the possibility of collisions when the construction is taking place near the arm which need quite a bit of manual work to avoid.
Freeze-dried Slow Salvos, Finally?
One slightly more expensive way to avoid all the manual work would be to use slmake to compile a slow salvo to build the whole Demonoid, or just the difficult part, from the far side of the construction-arm lane.

Compile slmake's output salvo with the freeze-dry script, and then compile the output of _that_ script with slmake. Then just trigger the seed constellation, and leave a long enough gap for the tricky construction to be completed. Easy! ... or at least a lot easier than all that clever stuff you put together by hand last time.

Unnecessary Tangent
It's probably worth looking at the true knightship Geminoid again, too. Being able to come up with easy self-destruct circuits with GoL-destroy, and then compile the result with slmake, makes these kinds of constructions really amazingly easy.

Which Circuitry, Though?
What were you thinking of as a base circuit for a 50%-reduced Demonoid, though? The old constructor-reflector has a repeat time of 153 -- seems like that could be improved, though I guess not at the cost of increasing the population or still-life count. Slower might even be fine if there's something really small.

Technically we'd only need a transparent output lane for the construction arm, not necessarily an edge shooter. In practice it seems hard to beat the NW31, though. With most other glider sources, a large number of 0hd recipes would have to be thrown away, because gliders couldn't be built close enough together.

When I try to build a candidate for half-0hd Demonoid circuitry, I keep ending up with big and kind of old-fashioned-looking stuff:

Code: Select all

x = 190, y = 225, rule = LifeHistory
122.3B$121.4B$120.4B$119.4B$118.4B$117.4B$117.4B$116.5B$113.9B$113.9B
$113.9B$113.10B$112.12B45.A$111.13B8.2A33.3A$112.11B9.A15.A17.A3B6.2A
3.2A$113.12B4.BA.A15.3A14.B2AB6.B2AB.B2AB$113.15B.B2A19.A12.4B4.B3.3B
2.2B$113.17B20.2A11.4B3.4B.3B.3B$113.17B20.B11.13B.7B5.2A$110.B.17B
19.3B9.23B5.A$109.2A19B16.6B4.8B.19B.BA.A$109.2AB.19B11.10B2.29B.B2A$
110.B4.20B2.2B3.44B$115.16BD15B2A3BD33B$115.9B.4BDBD15B2A2B2D33B$116.
7B2.4B3D18B2D33B$116.7B2.4BD21BD31B$116.6B4.26BD32B$116.7B10.4B.13B4.
25B3.2A$116.6B12.3B5.B.7B4.B4.20B3.A$115.7B13.4B15.2A7.15B6.3A$116.6B
15.2A15.A12.11B8.A$117.5B15.A17.3A8.13B$117.6B15.3A16.A7.15B$119.4B
17.A23.16B$118.B2A2B40.17B$119.2A43.16B$165.13B$165.5B2A2B.4B$167.3B
2A2B2.4B$167.8B2.B2AB$166.3B3D2B4.A3B$166.4BD3B5.3AB$166.2B3D2B7.BA2B
$166.7B8.4B$167.6B$167.6B$168.5B$168.6B$167.6B$167.7B$168.6B$168.6B$
168.6B$167.8B$166.8B$166.9B$166.9B$165.10B$165.3B2A5B$159.2A3.4B2A5B$
160.A3.11B$160.A.A7BD4B$161.2A2.4B3DB$166.2B2D2BD4.2A$167.6B4.A$167.
6B.BA.A$166.7B.B2A$167.8B$160.A6.7B$114.A45.3A4.7B$114.3A34.A11.A2.7B
4.B$101.2A3.2A6.3BA17.A15.3A8.2A3.6B.B.2BA$100.B2AB.B2AB6.B2AB14.3A
18.A7.15BA.A$101.2B2.3B3.B4.4B12.A20.2A3.B5.7BD4B.BA$102.3B.3B.4B3.4B
11.2A19.8B2.7B3D3B$94.2A5.7B.13B11.B21.15BD5B$95.A5.23B9.3B19.20B$95.
A.AB.19B.8B4.6B16.19B$96.2AB.29B2.10B11.21B$98.44B3.2B2.25B$98.33BD3B
2A15BD14B4.4B$98.33B2D2B2A15BDBD4B.6B6.4B$99.33B2D18B3D4B2.B.5B5.4B$
101.31BD21BD4B7.2A6.4B$99.32BD26B8.A8.4B$99.2A3.25B4.13B.4B16.3A6.4B$
100.A3.20B4.B4.7B.B5.3B19.A7.4B$97.3A6.15B7.2A17.2B29.4B$97.A8.11B12.
A15.2A32.4B$105.13B8.3A17.A33.2B$104.15B7.A16.3A35.B$104.16B23.A$103.
18B$28.3B73.16B$27.4B75.13B$26.4B75.4B.2B2A5B$25.4B75.4B2.2B2A3B$24.
4B76.2AB2.8B$23.4B75.3BA4.2B3D3B$23.4B74.B3A5.3BD4B$22.5B73.2BAB7.2B
3D2B$19.9B71.4B8.7B$19.9B75.A7.6B$19.9B74.A.A6.6B$19.10B73.A.A6.5B$
18.12B45.A24.3A.2A4.6B$17.13B8.2A33.3A23.A4.B6.6B$18.11B9.A15.A17.A3B
6.2A3.2A11.3AB2AB3.7B$19.12B4.BA.A15.3A14.B2AB6.B2AB.B2AB12.A.2AB.8B$
19.15B.B2A19.A12.4B4.B3.3B2.2B17.10B$19.17B20.2A11.4B3.4B.3B.3B18.6B
2A3B$19.17B20.B11.13B.7B5.2A10.6B2A2B5.2A$16.B.17B19.3B9.23B5.A11.10B
5.A$15.2A19B16.6B4.8B.19B.BA.A10.11B2.BA.A$15.2AB.19B11.10B2.29B.B2A
11.12B.B2A$16.B4.20B2.2B3.44B12.15B$21.16BD15B2A3BD33B11.16B$21.9B.4B
DBD15B2A2B2D33B8.2B.16B$22.7B2.4B3D18B2D33B8.2A18B$22.7B2.4BD21BD31B
10.2AB.17B$22.6B4.26BD32B9.B.4B.8B2.4B$22.7B10.4B.13B4.25B3.2A16.7B4.
4B$22.6B12.3B5.B.7B4.B4.20B3.A18.6B5.4B$21.7B13.4B15.2A7.15B6.3A17.4B
6.4B$22.6B15.2A15.A12.11B8.A19.3BA5.2B2D$23.5B15.A17.3A8.13B28.BA.A5.
BD.D$23.6B15.3A16.A7.15B28.A.A6.D$25.4B17.A23.16B29.A$24.B2A2B40.17B
30.3A$25.2A43.16B32.A$71.13B$71.5B2A2B.4B$73.3B2A2B2.4B$73.8B2.B2AB$
72.3B3D2B4.A3B$72.4BD3B5.3AB$72.2B3D2B8.A2B$72.7B8.4B$73.6B$73.6B$74.
5B$74.6B$73.6B$73.7B$74.6B$74.6B$74.6B$73.8B$72.8B$72.9B$72.9B$71.10B
$71.3B2A5B$65.2A3.4B2A5B$66.A3.11B$66.A.A7BD4B$67.2A2.4B3DB$72.2B2D2B
D4.2A$73.6B4.A$73.6B.BA.A$72.7B.B2A$73.8B$66.A6.7B$20.A45.3A4.7B$20.
3A34.A11.A2.7B4.B$7.2A3.2A6.3BA17.A15.3A8.2A3.6B.B.2BA$6.B2AB.B2AB6.B
2AB14.3A18.A7.15BA.A$7.2B2.3B3.B4.4B12.A20.2A3.B5.7BD4B.BA$8.3B.3B.4B
3.4B11.2A19.8B2.7B3D3B$2A5.7B.13B11.B21.15BD5B$.A5.23B9.3B19.20B$.A.A
B.19B.8B4.6B16.19B$2.2AB.29B2.10B11.21B$4.44B3.2B2.25B$4.33BD3B2A15BD
14B4.4B$4.33B2D2B2A15BDBD4B.6B6.4B$5.33B2D18B3D4B2.B.5B5.4B$7.31BD21B
D4B7.2A6.4B$5.32BD26B8.A8.4B$5.2A3.25B4.13B.4B16.3A6.4B$6.A3.20B4.B4.
7B.B5.3B19.A7.4B$3.3A6.15B7.2A17.2B29.4B$3.A8.11B12.A15.2A32.4B$11.
13B8.3A17.A33.2B$10.15B7.A16.3A35.B$10.16B23.A$9.18B$10.16B$12.13B$
11.4B.2B2A5B$10.4B2.2B2A3B$10.2AB2.8B$8.3BA4.2B3D3B$7.B3A5.3BD4B$6.2B
AB7.2B3D2B$5.4B8.7B$9.A7.6B$8.A.A6.6B$8.A.A6.5B$6.3A.2A4.6B$5.A4.B6.
6B$6.3AB2AB3.7B$8.A.2AB.8B$12.10B$12.6B2A3B$12.6B2A2B5.2A$12.10B5.A$
11.11B2.BA.A$11.12B.B2A$10.15B$9.16B$6.2B.16B$5.2A18B$5.2AB.17B$6.B.
4B.8B2.4B$13.7B4.4B$14.6B5.4B$16.4B6.4B$18.3BA5.2B2D$19.BA.A5.BD.D$
20.A.A6.D$21.A$22.3A$24.A!
It's certainly possible to go smaller, and even be HashLife-friendly while we're at it:

Code: Select all

x = 282, y = 169, rule = LifeHistory
243.2A$243.A.A$245.A4.2A$241.4A.2A2.A2.A$241.A2.A.A.A.A.2A$243.BABABA
.A$244.B2ABA.A$245.2B.BA$244.3B$235.2A6.4B2.4B$236.A6.B2A6B$236.A.AB
3.B2A5B$237.2AB.10B$239.13B$239.14B$239.15B$241.8B2.4B$241.6B5.4B$
240.9B4.3B$239.4B4.2A5.2B$238.4B5.A7.B$237.4B7.3A$236.4B10.A$134.A
100.4B$134.3A97.4B$77.3B57.A95.4B$76.4B56.2A94.4B$75.4B57.5B8.A81.4B$
74.4B60.4B5.3A80.4B$73.4B59.7B3.A82.4B$72.4B59.9B2.2A80.4B$72.4B18.A
38.11B.3B2.B10.B4.A11.2A47.4B$71.5B18.3A36.13B3.3B8.2B3.A.A9.B2AB45.
4B$68.9B20.A22.A11.2A18B7.3B3.A.A9.3B45.4B$68.9B19.2A20.3A11.2A18B6.B
2AB.3A.2A9.B.B43.4B$68.9B19.4B17.A15.B.18B4.2BAB.A4.B8.5B42.4B$68.10B
20.3B16.2A17.18B2.BABA3.3AB2A6.6B41.4B$67.12B18.4B33.20B.2B2A6.A.2A4.
8B40.4B$66.13B8.2A8.5B12.B7.B.7B2.25B9.13B40.4B$67.11B9.A8.6B11.2B.B.
13B2.24B12.13B37.4B$68.12B4.BA.A8.8B2.26BD23B12.15B35.4B$68.15B.B2A8.
14BD21BD23B2.B10.15B34.4B$68.17B11.13B3D18B2D28B5.B.17B32.4B$68.17B
12.7B.4BDBD15B2A2B2D51B31.4B$65.B.17B12.15BD15B2A3BD22B2ABD11B2A13B
29.4B$64.2A19B11.19B2.2B3.11B.21B2AB3D9B2A14B27.4B$64.2AB.19B8.17B11.
10B2.23BDBD20B3.B2A25.4B$65.B4.20B2.2B.15B16.6B3.25BD19B4.A2.A23.4B$
70.16BD22B19.3B5.28B2.2B2.B3.6B5.2A.A21.4B$70.9B.4BDBD13B.8B21.B4.13B
.7B2.4B13.6B7.A20.4B$71.7B2.4B3D9B2A2B3.8B19.2A3.13B2.6B17.9B6.2A13.B
4.4B16.A11.2A$71.7B2.4BD11B2A2B5.B3.2A20.A5.11B3.3B19.2A4.4B19.3B2.4B
16.A.A9.B2AB$71.6B4.18B10.A18.3A8.9B4.B21.A5.4B10.2A5.9B17.A.A9.3B$
71.7B10.4B.3B.B12.3A15.A10.8B24.3A7.4B10.A4.9B12.2A2.3A.2A9.B.B$71.6B
12.3B20.A26.8B24.A10.4B9.A.AB.8B13.A2.A4.B8.5B$70.7B13.4B47.6B36.B3D
9.2AB.9B3.4B2.BA.A3.3AB2A6.6B$71.6B15.2A47.5B38.D3B10.11B2.5B2.B2A6.A
.2A4.8B$72.5B15.A50.3B39.D3B9.11B2.8B10.13B$72.6B15.3A90.4B8.21B12.
13B$74.4B17.A91.4B8.19B12.15B$73.B2A2B110.4B6.19B2.B10.15B$74.2A113.
4B3.26B5.B.17B$190.4B.51B$191.4B.21B2A13B2A13B$179.2A11.25B2A13B2A14B
$179.A.A11.50B3.B2A$181.A4.2A6.48B4.A2.A$177.4A.2A2.A2.A3.33B2.2B2.B
3.6B5.2A.A$177.A2.A.A.A.A.2A3.16B2.7B2.4B13.6B7.A$179.BABABA.A6.16B3.
6B17.9B6.2A$180.B2ABA.A7.14B5.3B19.2A4.4B$181.2B.BA11.9B.3B4.B21.A5.
4B$180.3B14.8B3.2A23.3A7.4B$171.2A6.4B15.4B6.A24.A10.4B$172.A6.B2A3B
15.4B5.3A33.4B$172.A.AB3.B2A3B17.2A7.A34.4B$173.2AB.10B15.A44.4B$175.
13B15.3A42.4B$175.14B16.A43.4B$175.15B60.4B$177.8B2.4B60.4B$177.6B5.
4B60.4B$176.9B4.4B60.4B$175.4B4.2A5.4B60.4B$174.4B5.A7.4B60.4B$173.4B
7.3A5.4B60.4B$172.4B10.A6.4B60.4B$70.A100.4B19.4B60.4B$70.3A97.4B21.
4B60.4B$13.3B57.A95.4B23.4B60.4B$12.4B56.2A94.4B25.4B60.4B$11.4B57.5B
8.A81.4B27.4B60.4B$10.4B60.4B5.3A80.4B29.4B60.4B$9.4B59.7B3.A82.4B31.
4B60.4B$8.4B59.9B2.2A80.4B33.4B60.4B$8.4B18.A38.11B.3B2.B10.B4.A11.2A
47.4B35.4B60.4B$7.5B18.3A36.13B3.3B8.2B3.A.A9.B2AB45.4B37.4B60.4B$4.
9B20.A22.A11.2A18B7.3B3.A.A9.3B45.4B39.4B60.4B$4.9B19.2A20.3A11.2A18B
6.B2AB.3A.2A9.B.B43.4B41.4B60.4B$4.9B19.4B17.A15.B.18B4.2BAB.A4.B8.5B
42.4B43.4B60.4B$4.10B20.3B16.2A17.18B2.BABA3.3AB2A6.6B41.4B45.4B60.4B
$3.12B18.4B33.20B.2B2A6.A.2A4.8B40.4B47.4B60.4B$2.13B8.2A8.5B12.B7.B.
7B2.25B9.13B40.4B49.4B60.4B$3.11B9.A8.6B11.2B.B.13B2.24B12.13B37.4B
51.4B60.4B$4.12B4.BA.A8.8B2.26BD23B12.15B35.4B53.4B60.4B$4.15B.B2A8.
14BD21BD23B2.B10.15B34.4B55.4B60.4B$4.17B11.13B3D18B2D28B5.B.17B32.4B
57.4B60.4B$4.17B12.7B.4BDBD15B2A2B2D51B31.4B59.4B60.4B$.B.17B12.15BD
15B2A3BD22B2ABD11B2A13B29.4B61.4B60.3B$2A19B11.19B2.2B3.11B.21B2AB3D
9B2A14B27.4B63.4B60.2B$2AB.19B8.17B11.10B2.23BDBD20B3.B2A25.4B65.4B
60.B$.B4.20B2.2B.15B16.6B3.25BD19B4.A2.A23.4B67.4B$6.16BD22B19.3B5.
28B2.2B2.B3.6B5.2A.A21.4B69.4B$6.9B.4BDBD13B.8B21.B4.13B.7B2.4B13.6B
7.A20.4B71.4B$7.7B2.4B3D9B2A2B3.8B19.2A3.13B2.6B17.9B6.2A13.B4.4B16.A
11.2A43.4B$7.7B2.4BD11B2A2B5.B3.2A20.A5.11B3.3B19.2A4.4B19.3B2.4B16.A
.A9.B2AB43.4B$7.6B4.18B10.A18.3A8.9B4.B21.A5.4B10.2A5.9B17.A.A9.3B45.
4B$7.7B10.4B.3B.B12.3A15.A10.8B24.3A7.4B10.A4.9B12.2A2.3A.2A9.B.B45.
4B$7.6B12.3B20.A26.8B24.A10.4B9.A.AB.8B13.A2.A4.B8.5B46.4B$6.7B13.4B
47.6B36.B3D9.2AB.9B3.4B2.BA.A3.3AB2A6.6B47.4B$7.6B15.2A47.5B38.D3B10.
11B2.5B2.B2A6.A.2A4.8B48.4B$8.5B15.A50.3B39.D3B9.11B2.8B10.13B50.4B$
8.6B15.3A90.4B8.21B12.13B49.4B$10.4B17.A91.4B8.19B12.15B49.4B$9.B2A2B
110.4B6.19B2.B10.15B50.4B$10.2A113.4B3.26B5.B.17B50.4B$126.4B.51B51.
4B$127.4B.21B2A13B2A13B51.4B$128.25B2A13B2A14B51.4B$129.50B3.B2A51.4B
$130.48B4.A2.A51.4B$129.33B2.2B2.B3.6B5.2A.A51.4B$129.16B2.7B2.4B13.
6B7.A52.4B$129.16B3.6B17.9B6.2A52.4B$130.14B5.3B19.2A4.4B60.4B$133.9B
.3B4.B21.A5.4B60.4B$133.8B3.2A23.3A7.4B60.4B$134.4B6.A24.A10.4B60.4B$
136.4B5.3A33.4B60.3B$138.2A7.A34.4B60.2B$138.A44.4B60.B$139.3A42.4B$
141.A43.4B$186.4B$187.4B$188.4B$189.4B$190.4B$191.4B$192.4B$193.4B$
194.4B$195.4B$196.4B$197.4B$198.4B$199.4B$200.4B$201.4B$202.4B$203.4B
$204.4B$205.4B$206.4B$207.4B$208.4B$209.2B2A$210.BA.A$211.A3B!
or even

Code: Select all

x = 196, y = 210, rule = LifeHistory
77.3B$76.4B$75.4B$74.4B$73.4B$72.4B$72.4B$71.5B$68.9B$68.9B$68.9B$68.
10B$67.12B45.A$66.13B8.2A33.3A$67.11B9.A15.A17.A3B6.2A3.2A$68.12B4.BA
.A15.3A14.B2AB6.B2AB.B2AB$68.15B.B2A19.A12.4B4.B3.3B2.2B$68.17B20.2A
11.4B3.4B.3B.3B$68.17B20.B11.13B.7B5.2A$65.B.17B19.3B9.23B5.A$64.2A
19B16.6B4.8B.19B.BA.A$64.2AB.19B11.10B2.29B.B2A$65.B4.20B2.2B3.44B$
70.16BD15B2A3BD33B$70.9B.4BDBD15B2A2B2D33B$71.7B2.4B3D18B2D33B$71.7B
2.4BD21BD31B$71.6B4.26BD32B$71.7B10.4B.13B4.25B3.2A$71.6B12.3B5.B.7B
4.B4.20B3.A$70.7B13.4B15.2A7.15B6.3A$71.6B15.2A15.A12.11B8.A$72.5B15.
A17.3A8.13B$72.6B15.3A16.A7.15B$74.4B17.A23.16B$73.B2A2B40.18B$74.2A
43.16B$120.13B$120.5B2A2B.4B$122.3B2A2B2.4B$122.8B2.B2A$121.3B3D2B4.A
3B46.2A$121.4BD3B5.3AB45.A.A$121.2B3D2B7.BA2B46.A4.2A$121.7B8.4B41.4A
.2A2.A2.A$122.6B7.A45.A2.A.A.A.A.2A$122.6B6.A.A46.BABABA.A$123.5B6.A.
A47.B2ABA.A$123.6B4.2A.3A46.2B.BA$122.6B6.B4.A44.3B$122.7B3.B2AB3A36.
2A6.4B$123.8B.B2A.A39.A6.B2A3B$123.10B43.A.AB3.B2A3B$122.3B2A6B44.2AB
.10B$116.2A5.2B2A6B46.13B$117.A5.10B46.14B$117.A.AB2.11B45.15B$118.2A
B.12B47.8B2.4B$120.15B46.6B5.4B$120.16B44.9B4.3B$120.16B.2B40.4B4.2A
5.2B$120.18B2A38.4B5.A7.B$119.17B.B2A37.4B7.3A$118.4B2.8B.4B.B37.4B
10.A$13.3B101.4B4.7B43.4B$12.4B100.4B5.6B43.4B$11.4B100.4B6.4B44.4B$
10.4B100.2D2B5.A3B45.4B$9.4B100.D.DB5.A.AB45.4B$8.4B103.D6.A.A45.4B$
8.4B111.A45.4B$7.5B108.3A45.4B$4.9B84.2A10.4B7.A46.4B$4.9B85.A9.4B54.
4B$4.9B83.A10.4B54.4B$4.10B82.5A5.4B5.2A47.4B$3.12B45.A40.A4.4B5.A47.
4B$2.13B8.2A33.3A37.3AB2.7B.BA.A46.4B$3.11B9.A15.A17.A3B6.2A3.2A23.A.
2B3.7B.B2A46.4B$4.12B4.BA.A15.3A14.B2AB6.B2AB.B2AB22.4A12B47.4B$4.15B
.B2A19.A12.4B4.B3.3B2.2B21.2A2.BA3B2A7B46.4B$4.17B20.2A11.4B3.4B.3B.
3B21.A2.3AB.2B2A7B45.4B$4.17B20.B11.13B.7B5.2A13.2A.A.B3.10B44.4B$.B.
17B19.3B9.23B5.A17.A8.8B42.4B$2A19B16.6B4.8B.19B.BA.A17.2A7.9B40.4B$
2AB.19B11.10B2.29B.B2A28.3B2.4B38.4B$.B4.20B2.2B3.44B28.5B3.3B37.4B$
6.16BD15B2A3BD33B28.2A7.2B36.4B$6.9B.4BDBD15B2A2B2D33B29.A8.B35.4B$7.
7B2.4B3D18B2D33B27.3A44.4B$7.7B2.4BD21BD31B29.A45.4B$7.6B4.26BD32B67.
B4.4B16.A11.2A$7.7B10.4B.13B4.25B3.2A66.3B2.4B16.A.A9.B2AB$7.6B12.3B
5.B.7B4.B4.20B3.A59.2A5.9B17.A.A9.3B$6.7B13.4B15.2A7.15B6.3A57.A4.9B
12.2A2.3A.2A9.B.B$7.6B15.2A15.A12.11B8.A57.A.AB.8B13.A2.A4.B8.5B$8.5B
15.A17.3A8.13B56.B9.2AB.9B3.4B2.BA.A3.3AB2A6.6B$8.6B15.3A16.A7.15B55.
2B10.11B2.5B2.B2A6.A.2A4.8B$10.4B17.A23.16B55.3B9.11B2.8B10.13B$9.B2A
2B40.18B54.4B8.21B12.13B$10.2A43.16B56.4B8.19B12.15B$56.13B59.4B6.19B
2.B10.15B$56.5B2A2B.4B59.4B3.26B5.B.17B$58.3B2A2B2.4B59.4B.51B$58.8B
2.B2A60.4B.21B2A13B2A13B$57.3B3D2B4.A3B46.2A11.25B2A13B2A14B$57.4BD3B
5.3AB45.A.A11.50B3.B2A$57.2B3D2B7.BA2B46.A4.2A6.48B4.A2.A$57.7B8.4B
41.4A.2A2.A2.A3.33B2.2B2.B3.6B5.2A.A$58.6B7.A45.A2.A.A.A.A.2A3.16B2.
7B2.4B13.6B7.A$58.6B6.A.A46.BABABA.A6.16B3.6B17.9B6.2A$59.5B6.A.A47.B
2ABA.A7.14B5.3B19.2A4.4B$59.6B4.2A.3A46.2B.BA11.9B.3B4.B21.A5.4B$58.
6B6.B4.A44.3B14.8B3.2A23.3A7.4B$58.7B3.B2AB3A36.2A6.4B15.4B6.A24.A10.
4B$59.8B.B2A.A39.A6.B2A3B15.4B5.3A33.4B$59.10B43.A.AB3.B2A3B17.2A7.A
34.4B$58.3B2A6B44.2AB.10B15.A44.4B$52.2A5.2B2A6B46.13B15.3A42.4B$53.A
5.10B46.14B16.A43.4B$53.A.AB2.11B45.15B60.4B$54.2AB.12B47.8B2.4B60.4B
$56.15B46.6B5.4B60.4B$56.16B44.9B4.4B60.3B$56.16B.2B40.4B4.2A5.4B60.
2B$56.18B2A38.4B5.A7.4B60.B$55.17B.B2A37.4B7.3A5.4B$54.4B2.8B.4B.B37.
4B10.A6.4B$53.4B4.7B43.4B19.4B$52.4B5.6B43.4B21.4B$51.4B6.4B44.4B23.
4B$50.2D2B5.A3B45.4B25.4B$49.D.DB5.A.AB45.4B27.4B$51.D6.A.A45.4B29.4B
$59.A45.4B31.4B$56.3A45.4B33.4B$33.2A10.4B7.A46.4B35.4B$34.A9.4B54.4B
37.4B$32.A10.4B54.4B39.4B$32.5A5.4B5.2A47.4B41.4B$37.A4.4B5.A47.4B43.
4B$34.3AB2.7B.BA.A46.4B45.4B$33.A.2B3.7B.B2A46.4B47.4B$33.4A12B47.4B
49.4B$31.2A2.BA3B2A7B46.4B51.4B$30.A2.3AB.2B2A7B45.4B53.4B$30.2A.A.B
3.10B44.4B55.4B$33.A8.8B42.4B57.4B$33.2A7.9B40.4B59.4B$43.3B2.4B38.4B
61.4B$41.5B3.4B36.4B63.4B$41.2A7.4B34.4B65.4B$42.A8.4B32.4B67.4B$39.
3A10.3B31.4B69.4B$39.A13.2B30.4B71.4B$54.B24.B4.4B16.A11.2A43.4B$78.
3B2.4B16.A.A9.B2AB43.4B$70.2A5.9B17.A.A9.3B45.4B$71.A4.9B12.2A2.3A.2A
9.B.B45.4B$71.A.AB.8B13.A2.A4.B8.5B46.4B$72.2AB.9B3.4B2.BA.A3.3AB2A6.
6B47.4B$74.11B2.5B2.B2A6.A.2A4.8B48.4B$74.11B2.8B10.13B50.4B$65.B8.
21B12.13B49.4B$65.2B8.19B12.15B49.4B$65.3B6.19B2.B10.15B50.4B$65.4B3.
26B5.B.17B50.4B$66.4B.51B51.4B$67.4B.21B2A13B2A13B51.4B$68.25B2A13B2A
14B51.4B$69.50B3.B2A51.4B$70.48B4.A2.A51.4B$69.33B2.2B2.B3.6B5.2A.A
51.4B$69.16B2.7B2.4B13.6B7.A52.4B$69.16B3.6B17.9B6.2A52.4B$70.14B5.3B
19.2A4.4B60.4B$73.9B.3B4.B21.A5.4B60.4B$73.8B3.2A23.3A7.4B60.4B$74.4B
6.A24.A10.4B60.4B$76.4B5.3A33.4B60.3B$78.2A7.A34.4B60.2B$78.A44.4B60.
B$79.3A42.4B$81.A43.4B$126.4B$127.4B$128.4B$129.4B$130.4B$131.4B$132.
4B$133.4B$134.4B$135.4B$136.4B$137.4B$138.4B$139.4B$140.4B$141.4B$
142.4B$143.4B$144.4B$145.4B$146.4B$147.4B$148.4B$149.2B2A$150.BA.A$
151.A3B!
but it seems like the Snarks might get expensive, and so the smallest possible 0hd Demonoid won't have them. (?)

chris_c
Posts: 966
Joined: June 28th, 2014, 7:15 am

Re: Demonoid (diagonal Geminoid) completed!

Post by chris_c » October 21st, 2018, 9:46 am

dvgrn wrote: What were you thinking of as a base circuit for a 50%-reduced Demonoid, though? The old constructor-reflector has a repeat time of 153 -- seems like that could be improved, though I guess not at the cost of increasing the population or still-life count. Slower might even be fine if there's something really small.
I was thinking of using exactly the same constructor-reflector. I have tried but never been able to find anything obviously better. The geometry of the current 0hd-Demonoid is nice even though the repeat time is unsatisfyingly high. Like you say a bit later, it seems like Snarks are too expensive to appear in an "optimal" Demnoid. (By the way, my favourite definition of "optimal" is lowest period.)

EDIT: Maybe this one? But it still isn't obvious that the extra cost of construction is worth the benefit of going from repeat time 153 to 115.

Code: Select all

x = 71, y = 142, rule = B3/S23
23b2o$23bo15bo$21bobo15b3o$21b2o19bo$41b2o3$2o56b2o4b2o$2o55bo2bo3b2o$
58b2o$38b2o$38b2o5$63b2o$63b2o$28b2o$28bo$29b3o$31bo35b2o$10b2o55bo$
10b2o53bobo$65b2o3$51b2o$50bobo$50bo$49b2o7$54b2o$55bo$52b3o$52bo5$46b
o$46b3o$49bo$48b2o14b2o$64b2o12$65b2o$65bobo$67bo$67b2o5$49b2o$48bobo$
48bo$47b2o7$57b2o$57b2o7b2o$66bo$64bobo$64b2o$47b2o$47b2o$37b2o$38bo$
38bobo$39b2o9$43b2o$42bobo$42bo$41b2o6$58b2o$58b2o$50b2o$51bo$48b3o$
48bo2$49bo$48bobo$48bobo$46b3ob2o$45bo$46b3ob2o$48bob2o2$58b2o$58b2o7b
2o$67bo$65bobo$65b2o4$45b2o$45b2o5$61bo6b3o$60bobo5bo$60bobo6bo$61bo$
62b3o$64bo!

User avatar
dvgrn
Moderator
Posts: 10612
Joined: May 17th, 2009, 11:00 pm
Location: Madison, WI
Contact:

Re: Demonoid (diagonal Geminoid) completed!

Post by dvgrn » October 22nd, 2018, 5:57 pm

chris_c wrote:EDIT: Maybe this one? But it still isn't obvious that the extra cost of construction is worth the benefit of going from repeat time 153 to 115...
Ow. Population 177 instead of 126, but you can pack more elbow ops in, which will almost exactly make up for the expense.

126/177 = 0.712, but 115/153 = 0.752. Or if you go by number of still lifes instead of population, it's 21/28 = 0.75 exactly. That's a mighty close race as far as estimates go -- the only way to know for sure which one will come out smaller is to compile them both (which, fortunately, is not nearly as hard as it used to be).

However, there might be something within reach that's significantly smaller than either of those. 10hd construction arms are something like 25% more efficient that 0hd ones. Suppose we invented a seed constellation that converted a plain NW31 to one of those +10hd attachments?

We'd always build the NW31 version of the U.C., but each one would get magically converted to a +10hd attachment as soon as it moves from front position to back position, so the period would still be half of the old complicated 10hd model -- and we'd get the 25% efficiency advantage from using 10hd elbow recipes.

The idea of building a freeze-dried slow salvo for that part of the construction was already an option, to avoid the manual work that would otherwise be needed to build in the "danger zone". But probably the cost of the extra constellation would eat up pretty much all of the savings from using 10hd, so we might be back to a tossup again.

User avatar
calcyman
Moderator
Posts: 2932
Joined: June 1st, 2009, 4:32 pm

Re: Demonoid (diagonal Geminoid) completed!

Post by calcyman » October 27th, 2018, 5:37 pm

dvgrn wrote:This new faster HashLife-friendly Demonoid uses the same circuitry as Scorbie's Demonoid but has a somewhat higher population, because it has to spend a lot of gliders to build its child copies at a safer distance.

[...]

I'm kind of curious to see if a Demonoid with a 16384fd step size would need only something like 250 megabytes to run away. It would presumably overflow 2^21 so would have to be 2^22, with quite a bit higher population... at least unless a little more technology is developed -- at that distance it might be effective to build a 2-engine Cordership to get the elbow block out to the right distance.
That 'little more technology' has now been developed, and now it should be possible to have a HashLife-friendly Demonoid with a population only slightly larger than Scorbie's Demonoid, and to get speeds arbitrarily close to c/12.
What do you do with ill crystallographers? Take them to the mono-clinic!

User avatar
dvgrn
Moderator
Posts: 10612
Joined: May 17th, 2009, 11:00 pm
Location: Madison, WI
Contact:

Re: Demonoid (diagonal Geminoid) completed!

Post by dvgrn » November 13th, 2018, 6:10 pm

calcyman wrote:That 'little more technology' has now been developed, and now it should be possible to have a HashLife-friendly Demonoid with a population only slightly larger than Scorbie's Demonoid, and to get speeds arbitrarily close to c/12.
Here's a Demonoid puffer traveling at c/128, for starters -- "only '8' away from c/12!". This is what slsparse compiles for you automatically now when you give it a Demonoid infile.mc with a step size of 16384. The minimum power-of-two period with this particular recipe is 2^21.

-- Or rather, this is _almost_ what slsparse compiles. You do have to manually add in an elbow duplicator at one point, so that the recipe leaves behind an elbow at the farthest forward location, for the child U.C. to pick up and use without wasting time building a Cordership.

From here it's doable, though maybe not easy without some practice, to hand-edit this pattern using recipes from previous Demonoids, so that it uses just one Snarkbreaker instead of two Snarkmakers followed by two Snarkbreakers. The Scorbie Splitter should really be built with normal 90-degree gliders. Then the final Snark can be built with 90-degree gliders from a different direction, after the Snarkbreaker. slsparse doesn't know how to do things that way yet, and that's what's making the recipe so much more expensive than the Scorbie's Demonoid one (~119,000 cells total instead of ~72,000).

Then that can be followed by some version of the standard cleanup method with *WSSes -- but with 2-engine Corderships to push the new elbow out the required distances. I haven't quite thought of how to trick slsparse into producing the two Cordermakers for that part of the recipe, but no doubt it can be done somehow.

I _think_ that the current recipe is close enough to being able to run at 2^20 instead of 2^21, once the two unnecessary Snarkmaker recipes have been removed. Might even get back to fitting in a forum posting in macrocell format.
Attachments
lower-pop-Demonoid-puffer-2^21.mc
front end of a c/128 Demonoid
(322.42 KiB) Downloaded 521 times

User avatar
Goldtiger997
Posts: 762
Joined: June 21st, 2016, 8:00 am

Re: Demonoid (diagonal Geminoid) completed!

Post by Goldtiger997 » July 29th, 2020, 7:23 am

I made a fast Demonoid puffer using the recipe for 58P5H1V1 from this post. The veclocity is roughly (1,1)c/9.46 (I think the exact velocity is (4825784,4825784)c/45679544). It's not hashlife friendly, but I think it shouldn't be too hard to make it so by using shift.py.

(4825784,4825784)c_45679544 puffer.mc.gz
(1.06 MiB) Downloaded 241 times
I'm not sure what the best way to turn it into a working spaceship is. A glider could be fired back to trigger the destruction of the debris just before the faraway elbow block gets destroyed, but that glider would have to pass by the splitter, so unless the recipe is delayed (i.e. the spaceship is slowed down), we'd have to weave that backward-heading glider through the recipe stream coming out of the splitter.

User avatar
dvgrn
Moderator
Posts: 10612
Joined: May 17th, 2009, 11:00 pm
Location: Madison, WI
Contact:

Re: Demonoid (diagonal Geminoid) completed!

Post by dvgrn » July 29th, 2020, 9:04 am

Goldtiger997 wrote:
July 29th, 2020, 7:23 am
I made a fast Demonoid puffer using the recipe for 58P5H1V1 from this post. The [velocity] is roughly (1,1)c/9.46 ...
Wow, a Demonoid that can outrun a Cordership! How long a delay would have to be put in before it could outrun a seal?
Goldtiger997 wrote:
July 29th, 2020, 7:23 am
I'm not sure what the best way to turn it into a working spaceship is. A glider could be fired back to trigger the destruction of the debris just before the faraway elbow block gets destroyed, but that glider would have to pass by the splitter, so unless the recipe is delayed (i.e. the spaceship is slowed down), we'd have to weave that backward-heading glider through the recipe stream coming out of the splitter.
There are definitely lots of options here. We could design some self-destruct circuitry to go along with the Scorbie Splitter and the Snark, triggered by a glider that follows along after the single-channel recipe on a separate lane. The Scorbie Splitter self-destruct will send a glider to trigger the Snark self-destruct, which will put out a glider with exactly the right timing so that it gets to the opposite Scorbie Splitter self-destruct at exactly the right time, so the self-destruct signal neither catches up with the Demonoid nor falls behind it.

That's what the 10hd and 0hd Demonoids did, but that trick hasn't been used since.

A simpler alternative is to use the destruction method from more recent Demonoids. You wouldn't have to change your current fast Demonoid puffer at all, just make a cleanup recipe to follow along behind it -- consisting of

1) a Snarkbreaker (which requires adding a one-time turner to the construction of the c/5 spaceship seed, I think, on the southeast side -- build and then trigger a glider on the construction-arm Snark's output lane, and meet that with the rest of a Snarkbreaker recipe, back down at the parent Scorbie Splitter)
2) a huge pile of PUSH operations
3) a FIRE *WSS to knock out the grandparent Snark on that side
4) another huge pile of PUSH operations
5) a few more *WSSes to knock out the grandparent Scorbie Splitter.

The #5 single-channel recipe can be borrowed with no changes from previous Demonoids.

For #2 and #4, it will shorten the recipe quite a bit to build and shoot down two temporary Cordership. I think slsparse can be convinced to do this for you, if you give it something to compile that needs the right length of elbow push.

But it doesn't really matter how long the trailing cleanup recipe is, it won't affect the speed of the Demonoid or make you recompile anything. You'll just figure out a recipe and feed it in to the end of the puffer you've already made, and it will start cleaning it up.

... Honestly the 10hd/0hd "synchronized self-destruct signal" almost does seem simpler, though. Just needs a recompile for the additions to the Scorbie Splitter and Snark, and then you add one glider on the self-destruct lane, and it will just go along and do all the cleanup. That cleanup signal can be as close to the action or as far behind it as you want, technically making an unlimited number of "different" Demonoids... not interestingly different, just different.

User avatar
Moosey
Posts: 4306
Joined: January 27th, 2019, 5:54 pm
Location: here
Contact:

Re: Demonoid (diagonal Geminoid) completed!

Post by Moosey » July 29th, 2020, 12:35 pm

dvgrn wrote:
July 29th, 2020, 9:04 am
Goldtiger997 wrote:
July 29th, 2020, 7:23 am
I made a fast Demonoid puffer using the recipe for 58P5H1V1 from this post. The [velocity] is roughly (1,1)c/9.46 ...
Wow, a Demonoid that can outrun a Cordership! How long a delay would have to be put in before it could outrun a seal?
Well, now I have a naming proposal-- "Speed Demonoid"
not active here but active on discord

User avatar
calcyman
Moderator
Posts: 2932
Joined: June 1st, 2009, 4:32 pm

Re: Demonoid (diagonal Geminoid) completed!

Post by calcyman » July 29th, 2020, 5:53 pm

It might be worth mentioning some extra parts of the single-channel toolchain that were used in the construction of the 0E0P metacell.

Firstly, as well as outfile.mc, slmake emits an outfile.txt which contains the single-channel recipe in textual format. These text files can be concatenated using the standard command-line program cat, and this results in the recipes being concatenated.

If you don't have your glider stream in textual format, do not worry! There's a function destream() in lifelib which converts a pattern containing a glider stream into a Python list of integers -- https://gitlab.com/apgoucher/lifelib/-/ ... y#L643-659 -- as well as its inverse function, stream().

EDIT: The way that it's invoked is: glider_in_correct_orientation.destream(pattern_containing_glider_stream)

Secondly, lifelib/dd0e0p.cpp is a tool for inputting a tape (a textual glider recipe) into a pattern such as a Demonoid. It accepts an input file (text), a target file (an RLE or MC with exactly one glider* somewhere in the pattern, and still-lifes elsewhere), and an output file. The invocation:

Code: Select all

./dd0e0p if=recipe.txt tf=target.mc of=output.mc absolute=0 final
will conveniently replace the glider in the pattern file with the recipe (so the front of the recipe is where the original glider was).

*I think it works with arbitrary spaceships, so you can use it with an MWSS-stream Orthogonoid or a hypothetical loafer-stream spaceship, but in most cases you'll just want to have a glider stream.

Alternatively, you can use:

Code: Select all

./dd0e0p if=recipe.txt tf=target.mc of=output.mc relative=0 final
if you want the back of the recipe to be where the original glider was. This will perform the above substitution and then run the pattern for a number of generations equal to the length of the recipe (in order to 'load' the recipe tape into the Demonoid or whatever else is in target.mc). Here's the full list of documented options:

Code: Select all

/*
* dd0e0p -- write a logical glider stream to a physical pattern file.
*
* Mandatory options:
*
* if=stream.txt           Location of glider stream
* of=filename.mc          Pattern file to which to append
*
* Optional options:
*
* tf=target.mc            If you wish to not overwrite the target
* box=378,-43,3,3         Specify initial glider (rather than detect)
* final                   Remove initial glider afterwards
* absolute=523857         Advance pattern by n generations
* relative=1000           Advance pattern by n generations beyond recipe end
* bs=1048576              Advance by n generations at a time
*/
It runs the pattern in streamlife, which is faster than ordinary hashlife for patterns involving closely-crossing glider streams.
What do you do with ill crystallographers? Take them to the mono-clinic!

User avatar
Kazyan
Posts: 1247
Joined: February 6th, 2014, 11:02 pm

Re: Demonoid (diagonal Geminoid) completed!

Post by Kazyan » July 29th, 2020, 8:53 pm

Congratulations! This is such an interesting place for a c/5 diagonal ship.

A demonoid could probably become asymptotic to the the diagonal speed limit of c/4 as well. We could use the crab wickstretcher (synthesis from 2017 here), a clean ignition of the fuse, and a clean destruction like so:

Code: Select all

x = 68, y = 69, rule = B3/S23
59bo$59b2o$58bobo2$56b2o$56b2o2$58b2o6bo$57bo2bo5b2o$58bob2o3bobo$63bo
$64bo$60b2o2bo$59bo$58bobo$57bobo$56bobo$55bobo$54bobo$53bobo$52bobo6b
2o$51bobo8b2o$50bobo8bo$49bobo$48bobo$47bobo$46bobo$45bobo$44bobo$43bo
bo$42bobo$41bobo$40bobo$39bobo$38bobo$37bobo$36bobo$35bobo$34bobo$33bo
bo$32bobo$31bobo$30bobo$29bobo$28bobo$27bobo$26bobo$25bobo$24bobo$23bo
bo$22bobo$21bobo$20bobo$19bobo$18bobo$17bobo$16bobo$15bobo$14bobo$13bo
bo$12bobo$11bobo$10bobo$7b3obo$5bo4bo$4b2o3bo$3b2o3bo$6obo$2o3b2o!
Tanner Jacobi
Coldlander, a novel, available in paperback and as an ebook. Now on Amazon.

User avatar
Goldtiger997
Posts: 762
Joined: June 21st, 2016, 8:00 am

Re: Demonoid (diagonal Geminoid) completed!

Post by Goldtiger997 » July 30th, 2020, 5:39 am

Kazyan wrote:
July 29th, 2020, 8:53 pm
A demonoid could probably become asymptotic to the the diagonal speed limit of c/4 as well. We could use the crab wickstretcher (synthesis from 2017 here), a clean ignition of the fuse, and a clean destruction like so:...
Good idea! The more recent version of that synthesis takes 32 gliders — not too bad to create a seed for. It may be worth looking into some reductions of that recipe first though. The following looks like a pretty good way of igniting the fuse:

Code: Select all

x = 45, y = 53, rule = B3/S23
36bo$36b2o$35bobo2$33b2o$33b2o2$35b2o6bo$34bo2bo5b2o$35bob2o3bobo$40bo
$41bo$37b2o2bo$36bo$35bobo$34bobo$33bobo$32bobo$31bobo$30bobo$29bobo$
28bobo$27bobo$26bobo$25bobo$24bobo$23bobo$22bobo$21bobo$20bobo$19bobo$
18bobo$17bobo$16bobo$15bobo$14bobo$13bobo$12bobo$11bobo$10bobo$11bo$2o
$2o8$14b2o$13b2o$15bo!

User avatar
dvgrn
Moderator
Posts: 10612
Joined: May 17th, 2009, 11:00 pm
Location: Madison, WI
Contact:

Re: Demonoid (diagonal Geminoid) completed!

Post by dvgrn » July 30th, 2020, 8:43 am

Goldtiger997 wrote:
July 30th, 2020, 5:39 am
The following looks like a pretty good way of igniting the fuse...
Looks like the Doppler effect means that the fuse-lighting glider has the same effect every twelve ticks. Here's the same reaction, but with the glider coming from the direction it will probably be coming from, straight from the construction arm:

Code: Select all

x = 83, y = 84, rule = LifeHistory
73.2A$74.2A$73.A$70.2A$72.A2$70.A2.A$73.2A5.2A$75.A5.2A$73.A
.A4.A$74.A2.2A$77.2A$74.A$73.A.A$72.A.A$71.A.A$70.A.A$69.A.A
$68.A.A$67.A.A$66.A.A$65.A.A$64.A.A$63.A.A$62.A.A$61.A.A$60.
A.A$59.A.A$58.A.A$57.A.A$56.A.A$55.A.A$54.A.A$53.A.A$52.A.A$
51.A.A$50.A.A$49.A.A$48.A.A$47.A.A$46.A.A$45.A.A$44.A.A$43.A
.A$42.A.A$41.A.A$40.A.A$39.A.A$38.A.A$37.A.A$36.A.A$35.A.A$
34.A.A$33.A.A$32.A.A$31.A.A$30.A.A$29.A.A$28.A.A$27.A.A$26.A
.A$25.A.A$24.A.A$23.A.A$22.A.A$21.A.A$20.A.A$19.A.A$18.A.A$
17.A.A$16.A.A$15.A.A$14.A.A$13.A.A$12.A.A$11.A.A$10.A.A$2A7.
A.A$2A6.A.A$7.A.A$8.A$3A$2.A$.A!
-- I think it should be possible to convince slsparse to produce a recipe for the stretcher seed that doesn't require the extra two Snarkmakers/two Snarkbreakers trick to build the constellation from off to one side. With patterns as large as this I suppose the efficiency of the construction recipe doesn't really matter, but it shouldn't be too tricky to produce something, um, "relatively elegant".

These speed-Demonoids need two target blocks very far apart from each other. There's no way to bounce a following signal non-destructively off of one of these stretchers, as the last Demonoid cleverly did with the leading c/5 spaceship.

So I assume that the trick to use here would be to burn the fuse and create the elbow to build the farther-away reflector first (the singleton Snark) -- and then send a 180-degree glider backwards and leave a long gap before sending more gliders to collide with it and make a new elbow to build the Scorbie Splitter. There are single-channel recipes for doing all of that fairly efficiently, from some previous Demonoid project, I forget which one.

User avatar
Pavgran
Posts: 220
Joined: June 12th, 2019, 12:14 pm

Re: Demonoid (diagonal Geminoid) completed!

Post by Pavgran » July 30th, 2020, 12:33 pm

Isn't using wickstretcher less efficient?
For corderpush, every 64 tick delay pushes resulting block 16hd farther, so efficiency is 1/4 (hd/tick).
For the new c/5 push, every 4 ticks block is pushed 8 hd, so efficiency is 2.
For the newest wickstretcher push, every 3 ticks block is pushed 2 hd, so efficiency is only 2/3.

User avatar
dvgrn
Moderator
Posts: 10612
Joined: May 17th, 2009, 11:00 pm
Location: Madison, WI
Contact:

Re: Demonoid (diagonal Geminoid) completed!

Post by dvgrn » July 30th, 2020, 2:18 pm

Pavgran wrote:
July 30th, 2020, 12:33 pm
Isn't using wickstretcher less efficient?
For corderpush, every 64 tick delay pushes resulting block 16hd farther, so efficiency is 1/4 (hd/tick).
For the new c/5 push, every 4 ticks block is pushed 8 hd, so efficiency is 2.
For the newest wickstretcher push, every 3 ticks block is pushed 2 hd, so efficiency is only 2/3.
2/3 is bigger than 1/4, so wickstretchers are better than Corderships. Seems like there must be something wrong with the math for c/5, since 8hd every four ticks is basically lightspeed.

Anyway, it seems like the idea is that there's a range of speeds between c/5 and c/4 that can't be reached by chasing c/5 spaceships with gliders -- and anyway I think the wickstretcher-based Demonoid is going to to end up a lot smaller, because there are many fewer gliders to synchronize to get a wickstretcher seed.

User avatar
Pavgran
Posts: 220
Joined: June 12th, 2019, 12:14 pm

Re: Demonoid (diagonal Geminoid) completed!

Post by Pavgran » July 30th, 2020, 5:20 pm

dvgrn wrote:
July 30th, 2020, 2:18 pm
Seems like there must be something wrong with the math for c/5, since 8hd every four ticks is basically lightspeed.
The math is right. Compare these two situations:

Code: Select all

x = 54, y = 23, rule = LifeSuper
20.2M29.2M$20.2M29.2M$19.M2.M27.M2.M$16.2M.M2.M24.2M.M2.M$22.M30.M$
14.2M3.M2.M22.2M3.M2.M$10.2H2.2M5.M19.2H2.2M5.M$10.2H3.M.5M19.2H3.M.
5M$16.M30.M2$14.2H29.2H$13.M2G28.M2G$13.M30.M$11.2M29.2M$5.2M4.M24.2M
4.M$5.3M3.M5.3G16.3M3.M5.3H$3.M4.M8.G3D13.M4.M8.H3C$3.M3.M10.C15.M3.M
10.C$7.M11.D18.M11.C$2.2M.M.M25.2M.M.M$2M5.M23.2M5.M$2M4.2M23.2M4.2M$
2.4M27.4M!
The gliders are 4 ticks apart, the blocks are 8hd apart. So hd/ticks is 2. If the speed of ship is even closer to c/4, the efficiency will be more than 2.
dvgrn wrote:
July 30th, 2020, 8:43 am
Here's the same reaction, but with the glider coming from the direction it will probably be coming from, straight from the construction arm:

Code: Select all

x = 83, y = 84, rule = LifeHistory
73.2A$74.2A$73.A$70.2A$72.A2$70.A2.A$73.2A5.2A$75.A5.2A$73.A
.A4.A$74.A2.2A$77.2A$74.A$73.A.A$72.A.A$71.A.A$70.A.A$69.A.A
$68.A.A$67.A.A$66.A.A$65.A.A$64.A.A$63.A.A$62.A.A$61.A.A$60.
A.A$59.A.A$58.A.A$57.A.A$56.A.A$55.A.A$54.A.A$53.A.A$52.A.A$
51.A.A$50.A.A$49.A.A$48.A.A$47.A.A$46.A.A$45.A.A$44.A.A$43.A
.A$42.A.A$41.A.A$40.A.A$39.A.A$38.A.A$37.A.A$36.A.A$35.A.A$
34.A.A$33.A.A$32.A.A$31.A.A$30.A.A$29.A.A$28.A.A$27.A.A$26.A
.A$25.A.A$24.A.A$23.A.A$22.A.A$21.A.A$20.A.A$19.A.A$18.A.A$
17.A.A$16.A.A$15.A.A$14.A.A$13.A.A$12.A.A$11.A.A$10.A.A$2A7.
A.A$2A6.A.A$7.A.A$8.A$3A$2.A$.A!
To make that setup work, you would need to send debris-cleaning gliders and the rest of the recipe alongside the wick, not touching it, so that there would be no time wasted. If you just ignite the wick and send the remaining gliders after, then the wick will soon catch up with the ship and debris would wait there for a while, wasting time. The idea is that we want to begin new construction right after making an elbow, so the total construction time would be constant, and the travelling part would be as high as we wish.

(After making the previous post and the first version of this post, I've realised that efficiency in terms of hd/ticks doesn't matter here. It matters in diminishing total recipe length. And I've wrote that the setup above wouldn't work, but that was based on assumption that "straight from the construction arm" meant "from the same lane as the recipe")

User avatar
dvgrn
Moderator
Posts: 10612
Joined: May 17th, 2009, 11:00 pm
Location: Madison, WI
Contact:

Re: Demonoid (diagonal Geminoid) completed!

Post by dvgrn » July 30th, 2020, 6:17 pm

Pavgran wrote:
July 30th, 2020, 5:20 pm
(After making the previous post and the first version of this post, I've realised that efficiency in terms of hd/ticks doesn't matter here. It matters in diminishing total recipe length. And I've wrote that the setup above wouldn't work, but that was based on assumption that "straight from the construction arm" meant "from the same lane as the recipe")
Yeah, it's only "from the same elbow as the recipe (gliders)", and we can move that elbow to wherever we want it.

So we can send the entire cleanup and new-construction recipe before the wick is even triggered:

Code: Select all

x = 95, y = 96, rule = LifeHistory
85.2A$86.2A$85.A$82.2A$84.A2$82.A2.A$85.2A5.2A$87.A5.2A$85.A.A4.A$86.
A2.2A$89.2A$86.A$85.A.A$84.A.A$83.A.A$82.A.A$81.A.A$80.A.A$79.A.A$78.
A.A$77.A.A$76.A.A$75.A.A$74.A.A$73.A.A$72.A.A$71.A.A$70.A.A$69.A.A$
68.A.A$67.A.A$66.A.A$65.A.A$64.A.A$63.A.A$62.A.A$61.A.A$60.A.A$59.A.A
$58.A.A$57.A.A$56.A.A$55.A.A$54.A.A$53.A.A$52.A.A$51.A.A$50.A.A12.3A$
49.A.A15.A$48.A.A15.A$47.A.A$46.A.A$45.A.A$44.A.A13.3A$43.A.A16.A$42.
A.A16.A$41.A.A$40.A.A$39.A.A$38.A.A14.3A$37.A.A17.A$36.A.A17.A$35.A.A
$34.A.A$33.A.A$32.A.A15.3A$19.3A9.A.A18.A$21.A8.A.A18.A$20.A8.A.A$28.
A.A$27.A.A$26.A.A16.3A$25.A.A19.A$24.A.A19.A$23.A.A$22.A.A$12.2A7.A.A
$12.2A6.A.A$19.A.A$20.A12$.A$B2A$A.A$.B!
All those NWward gliders can be slow-salvo gliders -- they just look like that to keep the pattern small. The actual construction could be done off to the southeast of the (2,1)-pulled block.

The extra glider that I show being cleaned up here, on the northwest side of the wick, might even be useful in making a target for building the Scorbie Splitter at the halfway point, to avoid losing the time on that. Not sure exactly how that will work -- maybe hit that glider with a slow salvo that builds a one-time turner, and then at the right time trigger that to send a glider SE to interrupt a single-channel stream and make a new target?

(The idea would be to use a Snarkmaker back at the glider source, and convert to a single-channel stream as soon as the wickstretcher cleanup gliders have been sent. That way, the required Snark and Scorbie Splitter can be built at a reasonable offset to the southeast without being exponentially more expensive due to bending around two elbows.)

I don't think the expensiveness of the recipe really matters to this design's ability to travel faster than c/5, though -- a longer recipe would just mean a higher-period Demonoid. Seems like if you let the wickstretcher run long enough, the Demonoid will be able to go pretty fast. I'm not absolutely sure that the return-glider trick won't slow things down, though, maybe to a top speed below c/5 for all I know for sure yet.

EDIT: At worst we could build and trigger two wickstretcher seeds, right? Run them simultaneously, and then leave the elbow blocks in place to start the next seed construction from? Then the Scorbie Splitter and Snark constructions could happen more or less simultaneously, relative to when the tape gliders get there. (?)

User avatar
Kazyan
Posts: 1247
Joined: February 6th, 2014, 11:02 pm

Re: Demonoid (diagonal Geminoid) completed!

Post by Kazyan » July 31st, 2020, 2:21 pm

Here's a 21-glider (EDIT: 22) version of the wickstretcher, done via a method to add the wick to a crab that already exists. 20G (EDIT: 21) is very likely by finding a 1G cleanup:

Code: Select all

x = 219, y = 89, rule = B3/S23
192bo$192bobo$192b2o9$151bo$151bobo$151b2o4$116bo$114bobo$115b2o5$162b
o$162bobo$162b2o6$43bo$43b2o2b3o$42bobo2bo$48bo4$134bo$109bo23bobo$2bo
107b2o21bo2bo$obo42b3o61b2o17b3o3bobo$b2o3bobo126bo$6b2o98b2o$7bo42bo
54bobo25bo$50bo56bo25bo$50bo16b2o64bo$9b3o54b2o$9bo58bo$10bo$112bo$
112b2o$111bobo6$156b2o$156bobo$156bo$123bo$123b2o$122bobo3$110b2o84b3o
$109bobo78b2o4bo$100bo10bo78bobo4bo$100b2o88bo$99bobo3$120b2o$119bobo$
121bo5$206b2o$206bobo$206bo3$216b2o$216bobo$216bo!
The wick-adding component is here:

Code: Select all

x = 42, y = 38, rule = B3/S23
3bo$3b2o$2bobo2$2o$2o2$2b2o6bo$bo2bo5b2o$2bob2o3bobo$7bo$8bo$5bo2bo$
15bo$15bobo$15b2o$19b3o$13b2o4bo$13bobo4bo$13bo4$7b2o$6bobo$8bo5$29b2o
$29bobo$29bo3$39b2o$39bobo$39bo!
Since there's a glider trailing after the crab, you do have to be sure that the trailing glider fits with existing crab syntheses (as shown above). This is why a 3G LoM inserter is used instead of the 2G version--the 2G version would have a trailing glider so close to the crab that there's no space to actually put it there during the crab synthesis. It's kind of weird to think of gliders as stationary objects, but the reference frame (so to speak) is c/4 diagonal!

As an unrelated benefit, this allows for construction of the double wickstretcher, which I don't think has been done elsewhere:

Code: Select all

x = 106, y = 108, rule = B3/S23
o$b2o$2o8$5bo$6b2o$5b2o8$19bobo$20b2o$20bo2$64bo$62b2o$18bo44b2o$19b2o
$18b2o15$81bo$79b2o$80b2o11$30bo$31b2o$30b2o2$35bo$23bo11bobo$21bobo
11b2o$22b2o3$20bobo8bo$21b2o8bobo$21bo8bobo5bo$32bo3b2o$24b2o11b2o$23b
obo$25bo5bo$12b2o16bobo12bo$13b2o16b2o10b2o$12bo8b2o21b2o$22b2o7b2o$
21bo9b2o$29b2o7b2o$29b2o6b2o$39bo$35b2o$34bobo$36bo2$85bo$84b2o$78b2o
4bobo$77b2o$79bo4$32b2o$33b2o$32bo5$94b2o$93b2o$95bo3$104b2o$103b2o$
105bo!
EDIT: Miscounted gliders.
Tanner Jacobi
Coldlander, a novel, available in paperback and as an ebook. Now on Amazon.

Sarp
Posts: 221
Joined: March 1st, 2015, 1:28 pm

Re: Demonoid (diagonal Geminoid) completed!

Post by Sarp » July 31st, 2020, 2:27 pm

Wickstrecher in 21g:

Code: Select all

x = 209, y = 84, rule = B3/S23
208bo$206b2o$207b2o5$192bo$192bobo$192b2o9$151bo$151bobo$151b2o4$116bo
$114bobo$115b2o5$162bo$162bobo$162b2o6$43bo$43b2o2b3o$42bobo2bo$48bo4$
134bo$109bo23bobo$2bo107b2o21bo2bo$obo42b3o61b2o17b3o3bobo$b2o3bobo
126bo$6b2o98b2o$7bo42bo54bobo25bo$50bo56bo25bo$50bo16b2o64bo$9b3o54b2o
$9bo58bo$10bo$112bo$112b2o$111bobo6$156b2o$156bobo$156bo$123bo$123b2o$
122bobo3$110b2o84b3o$109bobo78b2o4bo$100bo10bo78bobo4bo$100b2o88bo$99b
obo3$120b2o$119bobo$121bo!
Edit: inherited the miscounting of gliders.
Last edited by Sarp on August 11th, 2020, 10:28 am, edited 2 times in total.
WADUFI

User avatar
dvgrn
Moderator
Posts: 10612
Joined: May 17th, 2009, 11:00 pm
Location: Madison, WI
Contact:

Re: Demonoid (diagonal Geminoid) completed!

Post by dvgrn » July 31st, 2020, 3:39 pm

Sarp wrote:
July 31st, 2020, 2:27 pm
Wic[k]strecher in 21g...
That's quite a reduction -- awesome!

As a small extra bonus, the long^N boat form of the wick can be ignited directly by a glider, instead of needing a honeyfarm explosion to kick things off:

Code: Select all

x = 59, y = 59, rule = B3/S23
49b3o$51bo$50bo$47b2o$47bo2$49b2o$49b2o5b3o$49bobo6bo$51b2o4bo$50bob4o
$53b2o$50bo$49bobo$48bobo$47bobo$46bobo$45bobo$44bobo$43bobo$42bobo$
41bobo$40bobo$39bobo$38bobo$37bobo$36bobo$35bobo$34bobo$33bobo$32bobo$
31bobo$30bobo$29bobo$28bobo$27bobo$26bobo$9b2o14bobo$8bobo13bobo$10bo
12bobo$22bobo$21bobo$20bobo$19bobo$18bobo$17bobo$16bobo$15bobo$14bobo$
13bobo$12bobo$12b2o2$24b3o$26bo$25bo$2o$b2o$o!
It will be interesting to convert this to a slow-salvo recipe. So far the only conversion of a big synthesis recipe to a seed has been Goldtiger997's c/5 spaceship seed, which makes 100 synchronized gliders that then produce the spaceship. For this new recipe it probably makes more sense to slow-salvo-construct a mango, then add a couple of blinkers, and then build two more sub-constellations to do the crab construction and then append the wickstretcher part.

Maybe the last two might as well just be one constellation, but it's nice that the two pieces can be moved diagonally apart from each other if that makes the construction easier.

User avatar
toroidalet
Posts: 1514
Joined: August 7th, 2016, 1:48 pm
Location: My computer
Contact:

Re: Demonoid (diagonal Geminoid) completed!

Post by toroidalet » July 31st, 2020, 3:44 pm

It's possible to clean the wick-adder up using a mango, which should make the seed smaller:

Code: Select all

x = 31, y = 26, rule = B3/S23
3bo$3b2o$2bobo2$2o$2o2$2b2o6bo$bo2bo5b2o$2bob2o3bobo$7bo$8bo$5bo2bo$
15bo$15bobo$15b2o12bo$19b3o6bobo$13b2o4bo7bo2bo$13bobo4bo6bobo$13bo14b
o4$7b2o$6bobo$8bo!
We might also be able to replace the 3 LOM gliders with a single glider from the southwest hitting a small constellation (using a beehive doesn't work; does anyone know of any other small LOM seeds?).
Any sufficiently advanced software is indistinguishable from malice.

User avatar
dvgrn
Moderator
Posts: 10612
Joined: May 17th, 2009, 11:00 pm
Location: Madison, WI
Contact:

Re: Demonoid (diagonal Geminoid) completed!

Post by dvgrn » November 8th, 2020, 9:52 am

The =https://www.conwaylife.com/forums/view ... eds thread and the Speed Demonoid thread picked up the discussion about building a crabsstretcher seed, and then the Speed Demonoid enabled by that seed. So I'll bring this thread back a previous topic and wrap up a minor loose end.

Way back not quite two years ago I posted the puffer part of lower-population, Relatively Fast Demonoid (i.e., not a Speed Demonoid, but at least it can approach an upper-bound speed of c/12). That was the hard part, but I never got around to going back and doing the easy part (the cleanup).

Here's a small step in the right direction:
lower-pop-Demonoid-2^22-predecessor.mc
lower-population HashLife-friendly Demonoid predecessor
(396.13 KiB) Downloaded 107 times
lower-pop-Demonoid-2^22.mc
completed p2^22 Demonoid from above predecessor
(463.14 KiB) Downloaded 114 times
This design is not really "lower-population" yet, for at least two reasons:

1) The recipe still uses slsparse's default construction to build the Scorbie Splitter, which is seriously suboptimal:
dvgrn wrote:
November 13th, 2018, 6:10 pm
The Scorbie Splitter should really be built with normal 90-degree gliders. Then the final Snark can be built with 90-degree gliders from a different direction, after the Snarkbreaker. slsparse doesn't know how to do things that way yet, and that's what's making the recipe so much more expensive than the Scorbie's Demonoid one (~119,000 cells total instead of ~72,000).
2) I deliberately appended a suboptimal cleanup method to the original puffer, to show the comparison. To push the destruction elbow out to where it needs to be to knock out the previous Snark, I just piled on a long string of PUSH44 recipes, just like in previous Demonoids. The second long push, to reach the location needed to knock out the previous Scorbie Splitter, is done "correctly", with a Corderpush. It's easy to see that the Corderpush is cheaper, both in terms of time required and the number of gliders needed.

Now, to really make an attempt at minimizing a Relatively Fast Demonoid's population, it seems like a more radical redesign will be needed:

3) We could rig up a synchronized self-destruct mechanism for the Scorbie Splitter and Snark, very much along the lines of what's in the Speed Demonoid. This would make the final Snarkbreaker and both of the long-distance push operations unnecessary. That might reduce the minimum HashLife-friendly period to just 2^20 -- or maybe even 2^19, not sure -- and the macrocell-format Demonoid pattern might be able to fit in a forum code block again.

Adjustability
If the suboptimal PUSH44s are replaced by a standard Cordership push, the Relatively Fast Demonoid predecessor will become trivially adjustable by just changing the spacing between various single-channel sub-streams (even if the Scorbie Splitter recipe isn't replaced).

Post Reply