17 in 17: Efficient 17-bit synthesis project

[Extrementhusiast]

Sure enough, #113 is probably doable, since a base can be made in 11G and there's a 4G budget for the spark in a relatively simple activation step:

Code: Select all

RLE

Yeah that's doable. 16G in total:

Code: Select all

RLE

That's actually #112, which is still good.

[Goldtiger997]

Here's a lead on #113 via a 14G synthesis of a related still-life:

Code: Select all

x = 116, y = 39, rule = B3/S23
27bo$27bobo$27b2o3$24bo$23bo$23b3o6$2bo$obo$b2o$10b2o48b2o48b2o$10bobo
47bo49bo$11b2o49bob2o46bob2o$61b2obo46b2obo$62bo2bo46bo2bo$25bo35bo2b
2o45bo2b2o$24b2o34bo46b2obo$24bobo33bobo43bobob2o$61b2o44bo$104b2o$
103bobo$105bo2$7bo61b2o$7b2o40b3o17bobo$6bobo10b2o30bo17bo$19bobo28bo
8b3o$19bo41bo$60bo2$21b2o$20b2o$22bo!

Note that if the ship is replaced by a snake then the reaction actually makes #113:

Code: Select all

x = 30, y = 10, rule = B3/S23
6b2o18b2o$5b3o17b3o$2o3b2ob2o15b2ob2o$obo3b4o9b2obo3b4o$b2o4b2o10bob2o
4b2o2$3b3o17b3o$bo3bo15bo3bo$bo19bo$b3o17b3o!

I've been trying to use this idea to solve #113 but I've been running into problems. Inserting the diagonal spark in 4G or less at generation 43 would solve it:

Code: Select all

x = 25, y = 34, rule = LifeHistory
22.A.A$22.2A$.A21.A$2.A2.A$3A3.2A14.A$5.2A14.A$21.3A2$4.A$2.A.A$3.2A
2$17.D$18.D11$9.A$9.2A$8.A.A10.2A$21.A.A$21.A3$23.2A$22.2A$24.A!

Alternatively if a 4G or 5G synthesis is found for the bottom junk that doesn't pass through where the snake is required to be then that should also solve #131. I was unable to find such a synthesis with popseq but maybe someone else with better tools can. Another option is synthesizing a modified version of that junk that evolves directly to produce a python rather than a long shillelagh. Possibly that method could solve #267 too.

[chris_c]

I've been trying to use this idea to solve #113 but I've been running into problems. Inserting the diagonal spark in 4G or less at generation 43 would solve it:

Code: Select all

x = 25, y = 34, rule = LifeHistory
22.A.A$22.2A$.A21.A$2.A2.A$3A3.2A14.A$5.2A14.A$21.3A2$4.A$2.A.A$3.2A
2$17.D$18.D11$9.A$9.2A$8.A.A10.2A$21.A.A$21.A3$23.2A$22.2A$24.A!

I found a suitable 3G reaction:

Code: Select all

x = 31, y = 28, rule = B3/S23
16bobo$16b2o$bo15bo$2bo2bo$3o3b2o8bo$5b2o8bo6bo$15b3o2b2o$21b2o$4bo$2b
obo18b2o$3b2o18bobo$23bo$28b3o$28bo$29bo4$9bo$9b2o$8bobo4b2o$15bobo$
15bo3$17b2o$16b2o$18bo!

[PHPBB12345]

#113 in 15G:

Code: Select all

x = 115, y = 28, rule = B3/S23
16bobo$16b2o$bo15bo$2bo2bo$3o3b2o8bo$5b2o8bo6bo$15b3o2b2o$21b2o$4bo$2b
obo18b2o33b2o48b2o$3b2o18bobo32bo2bob2o43bo2bob2o$23bo36b2obo46b2obo$
28b3o30bo2bo46bo2bo$28bo31bo2b2o45bo2b2o$29bo29bo46b2obo$59bobo43bobob
2o$60b2o44bo$103b2o$9bo92bobo$9b2o93bo$8bobo4b2o$15bobo50b2o$15bo32b3o
17bobo$50bo17bo$49bo8b3o$17b2o41bo$16b2o41bo$18bo!

[calcyman]

#113 in 15G:

Code: Select all

x = 115, y = 28, rule = B3/S23
16bobo$16b2o$bo15bo$2bo2bo$3o3b2o8bo$5b2o8bo6bo$15b3o2b2o$21b2o$4bo$2b
obo18b2o33b2o48b2o$3b2o18bobo32bo2bob2o43bo2bob2o$23bo36b2obo46b2obo$
28b3o30bo2bo46bo2bo$28bo31bo2b2o45bo2b2o$29bo29bo46b2obo$59bobo43bobob
2o$60b2o44bo$103b2o$9bo92bobo$9b2o93bo$8bobo4b2o$15bobo50b2o$15bo32b3o
17bobo$50bo17bo$49bo8b3o$17b2o41bo$16b2o41bo$18bo!

Well done!

The Shinjuku/Catagolue update process automatically discovered that this knocks the most expensive SL from 23 to 19 gliders:

Code: Select all

#C [[ GRID MAXGRIDSIZE 14 THEME Catagolue ]] 
#CSYNTH xs17_jhke0mqz1 costs 19 gliders (true).
#CLL state-numbering golly
x = 171, y = 29, rule = B3/S23
19bo$17b2o$18b2o$2bo$obo2bobo$b2o3b2o9bo$6bo10bobo2bobo$17b2o3b2o$
4bo18bo$5bo$3b3o65b2o44b2o45b2o$24b3o44bo2bob2o39bo2bob2o40bo2bob
2o$24bo48b2obo42b2obo35bo7b2obo$25bo4b2o42bo2bo42bo2bo35bo7bo2bo$
30bobo40bo2b2o41bo2b2o33b3o6bo2b2o$30bo41bo42b2obo46bo$72bobo39bob
ob2o45b2o$73b2o40bo$156bo4bo$111b3o40bobo3b2o$9b2o53b2o12b3o32bo
41b2o3bobo$10b2o51bobo12bo33bo55b2o$9bo6b3o46bo8b2o3bo87b2o$16bo
56bobo93bo$17bo57bo2$19bo$18b2o$18bobo!

It suffices, therefore, to reduce the sum of the last three steps to 5 gliders or fewer. Perhaps a very intensive collisrc run (aimed at the result of Chris's synthesis) could help here?

[Kazyan]

One extra glider and an Insertion of a dot spark at generation 32 would do it. To get around a backblast from Chris's inserter, this insertion would have to travel upwards at c/2 for a while:

Code: Select all

x = 35, y = 29, rule = LifeHistory
21.A$19.2A$20.2A$4.A$2.A.A2.A.A$3.2A3.2A9.A$8.A10.A.A2.A.A$19.2A3.2A$
6.A18.A$7.A$5.3A$26.3A$26.A$27.A4.2A$32.A.A$32.A3$12.D2$11.2A$12.2A$
11.A6.3A$2A16.A$.2A16.A$A$21.A$20.2A$20.A.A!

[chris_c]

I found some relevant 6G converters but can't seem to get down to 5G. There are only a few 2G reactions that act as required on the left hand side and I don't think any are clean. Therefore it requres finding a lucky 3G reaction on the top which will clean up everything. I might try again tomorrow.

Code: Select all

x = 115, y = 29, rule = B3/S23
114bo$112b2o$102bo10b2o$100bobo$101b2o4bo$48bobo54b2o$49b2o44bo10b2o$
49bo8bo34bobo$57bo6bo29b2o$9bobo2bobo12bo27b3o2b2o$5bo4b2o3b2o10b2o34b
2o$6bo3bo4bo12b2o15bobo$4b3o39b2o$46bo2$107b2o$106bobo$106bo$107bo2b2o
$18b2o39b2o47bo2bo$17bobo38bobo46b2obo$17bo40bo46bo2bob2o$18bo2b2o36bo
2b2o32bo8b2o$3o16bo2bo37bo2bo32b2o$2bo15b2obo37b2obo32bobo$bo14bo2bob
2o34bo2bob2o$7bo8b2o30bo8b2o$7b2o39b2o$6bobo38bobo!

Also there was this nice looking result from JLS but I couldn't use it to get a synthesis:

Code: Select all

x = 13, y = 10, rule = B3/S23
4bobo$2bob4o2$2bo8b2o$6b2o4bo$5bo2bobo$5b2o2b3o$b2o$obo5bob2o$2bo5b2ob
o!

EDIT: Direct 16G using Kazyan's idea:

Code: Select all

x = 47, y = 57, rule = B3/S23
33bo$31b2o$32b2o$4bo$2bobo2bobo$3b2o3b2o21bo$8bo22bobo2bobo$31b2o3b2o$
6bo30bo$7bo$5b3o13$38b3o$38bo$39bo4b2o$44bobo$44bo5$11b2o$12b2o$11bo
18b3o$2o28bo$b2o28bo$o$33bo$32b2o$32bobo9$19b3o$19bo$20bo19bo$13b2o24b
2o$14b2o23bobo$13bo6b2o$19b2o$21bo!

EDIT2: The upward c/2 travel is for 44 generations. I think that counts as a while :)

[Kazyan]

Phew, I asked for a long inserter and man-oh-man did you give me one. Good work!

#64 in 16G:

Code: Select all

x = 81, y = 42, rule = B3/S23
21bo$19bobo5b2o$bo18b2o4b2o$2bo25bo$3o17bo$20b2o$6bo12bobo$4bobo$5b2o
7$36bo$36bobo30b2o$36b2o31bo$71bo$70b2o$18b2o49bo$18bo2b2o45bo2b3o$19b
2obo46b2obo$74bo$73b2o2$76b2o$75bobo$76bo$b2o30b3o42b2o$2b2o29bo44bobo
$bo32bo43bo3$38bo$12b2o23b2o$11bobo23bobo$13bo2$34b2o$34bobo$34bo!

So the final still life is #293--just unexpected enough to be a satisfying dark horse.

Code: Select all

#C [[ THUMBNAIL ]]
x = 8, y = 8, rule = B3/S23
5bo$4bobo$4bo2bo$2b2obobo$2bo2b2o$3bo$3o$o!

[Freywa]

We should also add syntheses for the xs18s for which Shinjuku/Cata doesn't know any synthesis. They were completed in 2017, right?

Regarding The Last One, we could start with a hook-with-tail and bang on the remaining nine bits.

[Ian07]

We should also add syntheses for the xs18s for which Shinjuku/Cata doesn't know any synthesis. They were completed in 2017, right?

The wiki says they were all completed by November 2014 in this thread, though I've also heard that some of these syntheses were actually invalid to begin with. At some point I'm planning to work on adding some of the remaining synths from Mark's database as well as the forums and Discord.

[calcyman]

We should also add syntheses for the xs18s for which Shinjuku/Cata doesn't know any synthesis. They were completed in 2017, right?

The wiki says they were all completed by November 2014 in this thread, though I've also heard that some of these syntheses were actually invalid to begin with. At some point I'm planning to work on adding some of the remaining synths from Mark's database as well as the forums and Discord.

I also suspect that in many cases when people reduced xs18_A (which was on the unsolved list) to an xs18_B (not on the unsolved list), Mark only excluded xs18_B from the unsolved list because his expert system had reduced it to xs18_A (which is simpler in Mark's total order on the complexity of still-lifes). So lots of circular dependencies could have arisen due to misunderstanding what the 'unsolved list' actually meant.

Fortunately this is not an issue now that we have Shinjuku (and Catagolue to make progress easy to track).

[Kazyan]

#293 in 13G:

Code: Select all

x = 65, y = 25, rule = B3/S23
7bo$8bo46bobo$6b3o47b2o5bo$56bo4b2o$5bo56b2o$4b2o$4bobo$54b2o$43bo10bo
bo$bo7b3o32bo6bo3bobo$obo6bo32b3o5bobo3bo$o2bo6bo39bo2bo8b2o$b2o42bo5b
2o9bobo$45b2o15bo$44bobo$53b3o$55bo$47b3o4bo$49bo$48bo3$46b2o$47b2o$
46bo!

We're done.

[Sokwe]

Incredible! Congratulations to all of the participants!

In my first post tackling the original 17-bit synthesis project, I posted a 54 glider synthesis that I described as "obvious":

Code: Select all

x = 417, y = 58, rule = B3/S23
346bo$344bobo$345b2o4$299bo74bo$298bo74bo8bobo$298b3o72b3o6b2o$355bo
14bo12bo$37bo84bo233b2o11bo$38bo84bo231b2o12b3o$36b3o3bo78b3o239bo14bo
$32bo7b2o83bo238bo11b2o$33b2o6b2o82bobo234b3o12b2o$32b2o58bo32b2o5bo
235bo$91bo38b2o234bobo$29bo61b3o37b2o234b2o$30b2o3bo$29b2o5bo3b2o36bo
39bo39bo39bo39bo39bo39bo39bo$34b3o4bo33bobobob2o32bobobob2o32bobobo6bo
bo26bobobo35bobobo35bobobo35bobobo7bo27bobobo45bob2o$2o39bobo30bob2obo
bobo30bob2obobo32bob2obo6b2o26bob2obo34bob2obo34bob2obo34bob2obo8b2o
24bob2obo44bob2o2bo$b2o39b2o30bo4bo2bo31bo4bobo32bo4bob2o4bo26bo4bob2o
31bo4bob2o31bo4bob2o31bo4bob2o4b2o25bo4bob2o15bo25bo4b2o3bobo$o3b2o64b
o4b3obo35b3obob2o32b3obob2o7b3o22b3obobobo31b3obobo33b3obobo33b3obobo
33b3obobo5b2o7b2o27b3obo4b2o$4bobo61bobo6b2o38bobo37bobo10bo26bobo2bo
34bobobo35bobobo35bobobo35bobobo5bobo7b2o28bobo5bo$4bo64b2o47b2o15b2o
21b2o11bo26b2o38b2ob2o35b2ob2o36bobob2o3b3o28bobob2o3bo40bob2o$134b2o
30b2o150b2obo2bo3bo29b2obo2bo43b2ob2o$95b2o39bo28b2o35b3o117b2o5bo32b
2o51bo$26b2o67bobo32b3o34bo36bo2bo76b2o128b2o$27b2o46b2o12b2o4bo34bo
72bo2bo37b2o37bo2bo120b3o4bobo$26bo17b2o28bobo11b2o41bo74b3o30b3ob2o
38bo2bo120bo$43b2o31bo13bo150bo3bo38b2o118bo3bo$45bo32b3o159bo163b2o$
78bo189bo134bobo$79bo186bobo104b2o$267b2o31b2o70b2o$300bobo55b3o13bo$
300bo59bo$271b2o80b3o3bo$272bo82bo$231b2o39bobo79bo$232b2o39b2o12bo$
231bo3b2o51bo14bo$235bobo48b3o13b2o$235bo66bobo$287bo$287b2o$286bobo8$
260bo$260b2o$259bobo!

To think that it can now be done in 11! (thanks to Freywa)

Code: Select all

x = 118, y = 42, rule = B3/S23
38bo$36bobo$37b2o11$70bo3bobo$69bo4b2o$69b3o3bo$bo$bobo$b2o$69b2o
44bo$70b2o42bobo$obo66bo42b3obo$b2o108bo4b2o$bo109bob2o2bo$78bo33b
ob2o$3bo73bobo$3bobo72bo$3b2o$114b2o$114bobo$115bo$111b2o$57b2o51b
obo$56bobo53bo$58bo13b2o$71b2o$73bo3$60b2o$59bobo$61bo!

I must admit that in my vanity I went out of my way to make sure a few of my legacy components from the original project made it into the 17-in-17 project.

[Hdjensofjfnen]

Great work!
(It's satisfying to think that I chose correctly out of the final 5 - but it wasn't a hard choice, anyway.)

[simsim314]

Congrats everyone this is incredible!

[mniemiec]

Congratulations to everyone on another great achievement! Plus, a lot of the techniques developed here will also benefit syntheses in other areas as well.