Soup-based syntheses

[dvgrn]

A bit clean-up

Code: Select all

x = 33, y = 23, rule = B3/S23
12b2o$12b2o6$14b2o$14b2o5$31b2o$21bo9b2o$b2o17bobo$obo4b2o3b2o6bo2bo$o
bo4bobo2b2o7b2o$bo6bo$4b2o23bo$4b2o22bobo$28bobo$29bo!

Now it should be synthesised in 22 gliders

Adding still lifes is one of my old favorite ways to do cleanups, but unfortunately it's not usually cost-effective these days. A glider can take the place of that added block, which (maybe) saves one glider:

Code: Select all

x = 77, y = 53, rule = B3/S23
13bo$11bobo$12b2o8$obo$b2o$bo11$33b2o$33b2o5$50b2o$40bo9b2o$20b2o17bob
o$19bobo4b2o3b2o6bo2bo$19bobo4bobo2b2o7b2o$20bo6bo$23b2o23bo$23b2o22bo
bo$47bobo$48bo12$74b2o$74bobo$74bo!

It's only "maybe" because it's possible that the added block can be synthesized as part of a constellation at no extra cost. I haven't fired up the synthesise-patt and synthesise-constellation scripts to see what the options are for building multiple still lifes at once, but it looks like the final cost might end up being 16G-18G. Maybe a little less, if one of the practitioners of black-magic synthesis gets hold of the problem.

[GUYTU6J]

...A glider can take the place of that added block, which (maybe) saves one glider:

Code: Select all

x = 77, y = 53, rule = B3/S23
13bo$11bobo$12b2o8$obo$b2o$bo11$33b2o$33b2o5$50b2o$40bo9b2o$20b2o17bob
o$19bobo4b2o3b2o6bo2bo$19bobo4bobo2b2o7b2o$20bo6bo$23b2o23bo$23b2o22bo
bo$47bobo$48bo12$74b2o$74bobo$74bo!

...

xs20_8k4ria4z178c from 16G:

Code: Select all

x = 252, y = 53, rule = B3/S23
189bo$190bo$188b3o8$177bo$178b2o$177b2o3$108bo$109b2o$108b2o4$108bo$
109bo$107b3o39b2o58b2o$149b2o58b2o2$109b3o$2bo108bo$obo107bo138bo$b2o
83b2o38b2o38b2o58b2o20bobo$16bo29bo29bo9b2o28bo9b2o28bo9b2o21bo12bo13b
o9b2o19bo2bo$15bobo27bobo27bobo37bobo37bobo29bobo13bo11bobo25b2o2b2ob
2o$15bo2bo26bo2bo26bo2bo36bo2bo28b2o6bo2bo29b2o11b3o3b2o6bo2bo24bo5bo$
16b2o28b2o12bobo13b2o38b2o29b2o7b2o39b3o7b2o7b2o26b4obo$56b2o2b2o137bo
46bobo$55bobo3bo22bo39bo39bo33bo25bo$57bo25bobo37bobo37bobo57bobo$83bo
bo37bobo37bobo21b2o34bobo$9b3o52bo19bo39bo39bo23b2o34bo$9bo53b2o122bo
5b3o$10bo52bobo127bo$194bo9$249b3o$249bo$250bo!

For the record, it renders this 34G in Catagolue obsolete

Code: Select all

#C [[ GRID MAXGRIDSIZE 14 THEME Catagolue ]] 
#CSYNTH xs20_8k4ria4z178c costs 34 gliders (true).
#CLL state-numbering golly
x = 381, y = 27, rule = B3/S23
106bo112bo$107bo109bobo$105b3o23bo86b2o3bo$129b2o91bo$93bo27bo8b2o
90b3o105bo$91bobo26bo207b2o$92b2o26b3o206b2o$168bo45bo11bobo41bo
46bo50bo$obo163b3o43b3o11b2o40b3o44b3o48b3o$b2o36bo51bobo11bo59bo
3b2o40bo3b2o10bo39bo3b2o41bo3b2o45bo3b2o$bo3b3o30bobo51b2o10bobo
59bobo2bo40bobo2bo50bobo2bo41bobo2bo9b3o33bobo2bo$5bo33b2o51bo12b
2o60b2obobo40b2obobo50b2obo43b2obo10bo36b2obo$6bo163bobo43bobo53b
2o7bo37b2o10bo38b2o$170b2o44b2o54bo8bobo35bo50bo2bo4b2o$92b2o72b2o
105bo7b2o38bo6b2o42b2o4bobo$93b2o71b2o104b2o46b2o5b2o35bo13bo$92bo
34b2o200bo34b2o$126b2o35b2o198bobo3b2o$52bo75bo33bobo204bobo$52bob
o109bo204bo$52b2o60bo53b2o59b2o$113bobo51bo2bo57b2o32b2o64b3o$49b
2o62bobo52b2o36bo9b2o12bo30bobo4bo3b3o41b3o9bo$49bobo62bo57b2o32b
2o7b2o46bo3b2o3bo6b3o36bo10bo$49bo35b2o85bobo30bobo9bo6bo42bobo3bo
5bo37bo$86b2o84bo50b2o55bo$85bo137bobo!

By the way, xs20_gbbgf96z1221 from this reaction could probably be reduced to <9G:

Code: Select all

x = 13, y = 15, rule = B3/S23
bo$obo$bo$2b2o$3bo4$6bo$5bobo3b2o$5bobo3b2o$6bo$2bo$2bo$2bo!

Unfortunately neither of the 3G synthesises for the seed at top has enough clearance:

Code: Select all

x = 33, y = 11, rule = B3/S23
2bo24bo$obo3b2o20bo$b2o3bobo17b3o$6bo24bo$31b2o$30bobo2$9bo$8b2o18bo$
8bobo17b2o$27bobo!

[dvgrn]

By the way, xs20_gbbgf96z1221 from this reaction could probably be reduced to <9G:

Code: Select all

x = 13, y = 15, rule = B3/S23
bo$obo$bo$2b2o$3bo4$6bo$5bobo3b2o$5bobo3b2o$6bo$2bo$2bo$2bo!

Unfortunately neither of the 3G synthesises for the seed at top has enough clearance...

Well, that's not really a problem, though -- just have to build all the pieces at the same time. Funny how the beehive is randomly accidentally transparent:

Code: Select all

x = 32, y = 33, rule = LifeHistory
.BA$ABAB$.2A2B$2.4B$3.4B$4.4B4.2B$5.4B2.4B$6.4B.4B$7.4B.3B$8.BD6B$8.D
BD6B$8.BD8B$9.B2D7B$10.BD8B$9.12B$14.6B2A$15.5BABA$14.D5BAB$12.BDBD3B
2D2B$12.BDBD3B2D2B$13.BD8B$14.9BA$9.3C3.7B2AB$15.7BABAB$14.4B2.7B$13.
4B3.2B.5B$12.4B8.2B3A$11.4B10.BA2B$10.4B12.BA2B$9.2A2B14.4B$8.2B2A16.
B3A$9.AB18.A2B$30.AB!

There might well be a 4G recipe for all three pieces of bait junk, but I haven't gotten around to making 4G searches really trivial to run on my system, like 3G searches are.

(Where is the latest 4G script-and-database, again? And how much bigger could the 4G database usefully be made? Would it make sense for me to check in a copy next to the 3-glider scripts, so they're all in the same place?)

[wildmyron]

(Where is the latest 4G script-and-database, again? And how much bigger could the 4G database usefully be made? Would it make sense for me to check in a copy next to the 3-glider scripts, so they're all in the same place?)

Here: https://github.com/apaap/b3s23life/tree ... stellation
There hasn't been any change in some time. I got stuck while working on the details of filtering boring duplicate collisions from the data before significantly extending the set of 4G collisions to include.

If you like, i can send a pull request to merge my 4G collision branch in its current state to your master.

[dvgrn]

Here: https://github.com/apaap/b3s23life/tree ... stellation
There hasn't been any change in some time. I got stuck while working on the details of filtering boring duplicate collisions from the data before significantly extending the set of 4G collisions to include.

If you like, i can send a pull request to merge my 4G collision branch in its current state to your master.

I might just add a README for now with a link to your branch. Eventually it might make sense to create a separate "four-glider-collisions" folder with your version of the database and script. The 3G search is nice and fast, and is exactly what is wanted for a lot of quick easy searches. Seems like the 4G search will be more along the lines of "bringing out the big guns", where you might wait half an hour for an answer.

(And if you weren't careful to search the 3G space first, you might get a million recipes in your output.)

Might it work to convert all the glider recipes to Shinjuku format, and filter out duplicates that way?

[wildmyron]

Here: https://github.com/apaap/b3s23life/tree ... stellation
There hasn't been any change in some time. I got stuck while working on the details of filtering boring duplicate collisions from the data before significantly extending the set of 4G collisions to include.

If you like, i can send a pull request to merge my 4G collision branch in its current state to your master.

I might just add a README for now with a link to your branch. Eventually it might make sense to create a separate "four-glider-collisions" folder with your version of the database and script. The 3G search is nice and fast, and is exactly what is wanted for a lot of quick easy searches. Seems like the 4G search will be more along the lines of "bringing out the big guns", where you might wait half an hour for an answer.

Until now I've only implemented the 4G search for constellations, and it's still very fast. For simple targets it does produce a lot of results though, so only getting 3G results would be useful.

(And if you weren't careful to search the 3G space first, you might get a million recipes in your output.)

This is why the 3G and 4G collisions are in separate files - the 3G collisions are searched (and presented) first. Also I had discussed with Goldtiger some possible options of limiting the number of gliders / total number of results and some other possibilities - but that's still at idea stage.

Might it work to convert all the glider recipes to Shinjuku format, and filter out duplicates that way?

I've changed the format of files for the collisions search: there is now only one file containing constellations and collision patterns. The collisions have been converted to Shinjuku format during processing and this resulted in some deduplication of the 3G collision set. The new file format also uses the Shinjuku gliderset format to store the collisions.

Unfortunately that is not sufficient because even the limited 4G survey I ran to generate the dataset results in many collisions of the form (3G + G) or (2G + 2G) where the only difference is in arrival time of the last glider (or glider pair). My intended strategy to deal with this was to amend the format to allow collisions with a non-empty starting pattern and to adjust the 4G survey to detect when their was an intermediate stable object. This would by no means be foolproof but instead of 100 trivially different collisions their might be only 3 or 4. Some further discussion on the topic can be found in the #tools channel on the Discord server (from back in (early?) December last year).

Anyway, this should probably be on a different thread, but I don't think there's a dedicated topic for the 3G pattern/collision search scripts yet. I remember I spent quite a bit of time digging through the 3G collision enumeration thread while working on this idea.

Also, in case anyone wants to pick this up: The scripts used to generate the new 3G and the limited 4G dataset are here: https://gist.github.com/apaap/e942a7185 ... 453ea73366 and https://gist.github.com/apaap/d291a4327 ... dda4fd58b4 (Hopefully they were the latest versions, although I did play around with some of the parameters in variations of the processing stage.

[GUYTU6J]

9G xs26_6a8ohf0fho8a6

Code: Select all

x = 99, y = 92, rule = B3/S23
bo95bo$2bo93bo$3o93b3o51$52bo$51bo$51b3o2$46bo$47bo$45b3o5$51bobo5bobo
$51b2o6b2o$52bo7bo$49bo$50bo$48b3o17$58b3o$58bo$59bo$41b2o$40bobo$42bo!

The three gliders at the bottom are for converting a TL-predecessor (see gen 53) to pi, which may be improved.
EDIT: 5G xs23_39e0o4czc970123 improved from 8G, thanks to a 3G pi+boat given by synthesise_patt_v1.2.py:

Code: Select all

x = 25, y = 34, rule = B3/S23
6bo$4bobo$5b2o13$18bo$18b2o2b2o$17bobo2bobo$22bo5$11b2o$10bobo$12bo6$b
2o$obo$2bo!

EDIT2: clean up this for xs22_0c9baiczck871 please (also see viewtopic.php?f=2&t=4257#p87785)

Code: Select all

x = 22, y = 14, rule = B3/S23
4bo$5b2o$4b2o5$16bobo$16b2o$17bo$13bo$3o8bobo6b2o$2bo9b2o5b2o$bo19bo!

(Apologies for a newbie's random practice)

[GUYTU6J]

Here's a seed for xs46_4a9b8bb8b96z259d1dd1d96:

Code: Select all

x = 16, y = 29, rule = B3/S23
13b2o$12bo2bo$12bo2bo$13bobo2$4b2o$bo2b2o$2b3o$3bo3$o$o$o$5b3o$o$o$o3$
3bo$2b3o$bo2b2o$4b2o2$13bobo$12bo2bo$12bo2bo$13b2o!

I need to insert two R's at gen 93 here

Code: Select all

x = 49, y = 51, rule = B3/S23
45bo$43b2o$44b2o6$2bo$obo$b2o22$6bo$4bobo$5b2o2$20b3o$27bo$27bo$27bo$
20b3o7$27b2o3bo$26bobo2b2o13b3o$28bo2bobo12bo$47bo!

[dvgrn]

I need to insert two R's at gen 93 here

Code: Select all

x = 49, y = 51, rule = B3/S23
45bo$43b2o$44b2o6$2bo$obo$b2o22$6bo$4bobo$5b2o2$20b3o$27bo$27bo$27bo$
20b3o7$27b2o3bo$26bobo2b2o13b3o$28bo2bobo12bo$47bo!

It looks like a lot of other things would work instead of the R's, such as various B-heptomino-shaped things that the R produces. This at T=105 is fairly clean, for example:

Code: Select all

x = 11, y = 29, rule = B3/S23
10bo$9bo$8b2o$9b2o$10bo$6b2o$2b2o2bobo$bo6b2o$b3o4bo$bo2b2obo$3ob2o$o
3b2o$o3bo$o3bo$6b3o$o3bo$o3bo$o3b2o$3ob2o$bo2b2obo$b3o4bo$bo6b2o$2b2o
2bobo$6b2o$10bo$9b2o$8b2o$9bo$10bo!

Even better, there are ways to replace the R with a couple of gliders:

Code: Select all

x = 19, y = 37, rule = B3/S23
18bo$16b2o$17b2o6$14bo$6b2o4b2o$2b2o2bobo4b2o$bo6b2o$b3o4bo$bo2b2obo$
3ob2o$o3b2o$o3bo$o3bo$6b3o$o3bo$o3bo$o3b2o$3ob2o$bo2b2obo$b3o4bo$bo6b
2o$2b2o2bobo4b2o$6b2o4b2o$14bo6$17b2o$16b2o$18bo!

I think there's even a way to replace each R with a single glider, but it might take a little hunting around in Seeds of Destruction to find it.

[GUYTU6J]

Here's the seed for that very rare xs42:

Code: Select all

x = 17, y = 23, rule = B3/S23
b2o$b2o11bobo$14b2o$4b2o9bo$3bo2bo$4b2o3$3o9bo$10b2ob2o$12bo2$12bo$10b
2ob2o$3o9bo3$4b2o$3bo2bo$4b2o9bo$14b2o$b2o11bobo$b2o!

Unluckily I didn't find a way to replace the blocks with gliders, and failed to settle the block+beehive:

Code: Select all

x = 45, y = 43, rule = LifeHistory
42.A.A$42.2A$43.A7$38.A$18.2A18.A.A$18.2A18.2A2$21.2A$A19.A2.A$.2A18.
2A$2A11$4.2A15.2D$5.2A13.D2.D5.A$4.A16.2D3.A2.A.A$24.A.A2.2A$18.2D5.
2A$.2A15.2D$A.A$2.A6$43.A$42.2A$42.A.A!

EDIT: better clean up, anyone?

Code: Select all

x = 303, y = 213, rule = B3/S23
295bobo$273bo21b2o$272bo23bo$272b3o3$276bo$275bo$275b3o21$278bo$278bob
o$278b2o11$14bo$15bo$13b3o11$81bo$80bo$80b3o2$113bo$113bobo$113b2o8$
76bo$74b2o$75b2o6$71b2o$70b2o$66bo5bo$61b2o2b2o$60bobo2bobo$62bo$44bo$
45bo$43b3o3$164bo2$164bo2$43bo22b2o96bo$43b2o20b2o$42bobo22bo96bo2$
164bo2$164bo2$164bo2$164bo$211bo$164bo44b2o$210b2o$164bo39bo$205bo$
164bo38b3o2$164bo24b2o$188bobo14b3o$164bo25bo16bo$206bo$164bo2$164bo
10$110b2o$110bobo$110bo34$113b2o$b2o110bobo$obo110bo$2bo8$260b2o$260bo
bo$120b2o138bo$120bobo$120bo9$293b3o$293bo$294bo12$278b2o$278bobo19b2o
$278bo21bobo$300bo4$248b2o$248bobo$248bo$274b2o$274bobo$274bo!

[dvgrn]

Here's the seed for that very rare xs42...
Unluckily I didn't find a way to replace the blocks with gliders, and failed to settle the block+beehive:

Code: Select all

x = 45, y = 43, rule = LifeHistory
42.A.A$42.2A$43.A7$38.A$18.2A18.A.A$18.2A18.2A2$21.2A$A19.A2.A$.2A18.
2A$2A11$4.2A15.2D$5.2A13.D2.D5.A$4.A16.2D3.A2.A.A$24.A.A2.2A$18.2D5.
2A$.2A15.2D$A.A$2.A6$43.A$42.2A$42.A.A!

That's fixable. There's only one 3G option for the active reaction on the right, but it's made up of a 2G TL+glider collision that makes an annoying spark, plus a third glider. The 2G can be replaced by three gliders that make the TL+glider predecessor without the annoying spark.

Once that's done there's all kinds of room to build everything that's needed. The above pattern also builds the blinkers in the wrong phase, which might have been causing some of the trouble.

Code: Select all

x = 45, y = 57, rule = LifeHistory
27.BD$26.2DB$25.2B2D$24.4B$23.4B$22.4B$21.4B$20.4BD17.A.A$5.DBD7.3B.
4BD3B15.2A$5.B2DB5.9B3D17.A$6.D3B3.12B$7.4B3.12B$8.4B.14B$9.18B$9.17B
$7.19B$7.19B12.A$8.10B2A6B12.A.A$10.8B2A6B12.2A$14.10B$14.3B2.2B2AB$.
A18.A2BA$2.A18.2A$3A31.E$34.E.E$34.2E$20.E$21.E$19.3E14.E$34.2E$35.2E
4$4.3A14.2D$6.A13.D2BD$5.A8.3B2.2B2DB$14.10B$10.8B2D6B$.2A5.10B2D6B$
2.2A3.19B$.A5.19B$9.17B$9.18B$8.4B.14B$7.4B3.12B$6.C3B3.12B$5.B2CB5.
9B3C17.A$5.CBC7.3B.4BC3B15.2A$20.4BC17.A.A$21.4B$22.4B$23.4B$24.4B$
25.2B2C$26.2CB$27.BC!

[dvgrn]

EDIT: better clean up, anyone?

Yeah, seven cleanup gliders is a bit excessive, according to Seeds of Destruction. This brings the upper bound down to four:

Code: Select all

x = 79, y = 99, rule = B3/S23
obo$b2o$bo3$58bo$57bo$57b3o8$76bo$76bobo$76b2o2$53bo$51b2o$52b2o6$48b
2o$47b2o$43bo5bo$38b2o2b2o$37bobo2bobo$39bo$21bo$22bo$20b3o7$20bo22b2o
$20b2o20b2o$19bobo22bo52$73b2o$73bobo$73bo!

[Kazyan]

This oscillator now has a soup, as of a few days ago. The rotor is currently unique in C1. Perhaps it's useful. https://catagolue.appspot.com/object/xp ... c826/b3s23

[GUYTU6J]

This oscillator now has a soup, as of a few days ago. The rotor is currently unique in C1. Perhaps it's useful. https://catagolue.appspot.com/object/xp ... c826/b3s23

The soup seems too active there:

Code: Select all

x = 76, y = 32, rule = B3/S23
3o14bo$16b3o$4bo6b3ob2o2bo53b2o$3bobo9b5o50bob2o$3bobo3bo5b3o52b3o$4bo
4bo6b2o53bo$9bo7b2o$18bo55bo$11b3o57b5o$65b2o3b3o2bo$66bo3b3o2$68bobo$
70bo$12b2obob2o48bo2bo$12b2obo3bo48bo$11bo7bo3b3o42bo$12b3o3b2o$13b5o$
6b2o7b3o$5bo2bo$5bobo$6b2o3$8b2o$8b2o$7bo$6b3o$7bo$8bob3o$10b3o!

EDIT: for today's random synthesis practice, 15G xs37_0g88baacz34311dlhe0eic:

Code: Select all

x = 312, y = 122, rule = B3/S23
195bo$193bobo$194b2o29$215bo$213bobo$214b2o3$100bo$100bobo$100b2o2$
129bo$130bo$128b3o122bo$252bobo$252bobo$140bobo110bo$141b2o3bo$141bo4b
obo108b2o$72b3o71b2o4b3o101bo2bo2b3o$257b2o2$253bo$252bobo$252bobo$
234bobo16bo$235b2o$obo155b2o75bo32b2o$b2o74bo80b2o81b3o24b2o$bo3bobo
67b2o88bo75bo33bo$5b2o69b2o86bobo75bo31bobo$6bo60bo79bo15bo2bo71bo18bo
15bo2bo$67bo79bo16b2o72b2o17bo16b2o$67bo8b2o69bo89bobo17bo$76bobo$76bo
4$35bo$34b2o$34bobo49$310b2o$309b2o$311bo!

There's another reason why I post it here: the beehive-to-R-bee reaction in the last few generations looks like a component. But how to make it a real converting reaction?

[Ian07]

A zz_linear growth with period 6528 has a surprisingly simple recipe:

Code: Select all

x = 22, y = 11, rule = B3/S23
2bo16bo$2ob2o13bobo$2bo14bo3bo$18bobo$19bo2$9b2o$9bobo$11bo8b2o$9bobo
8b2o$9b2o!

Haul: https://catagolue.appspot.com/haul/b3s2 ... 2074c774e2 (Rob Liston, 2020-01-22)

[GUYTU6J]

A zz_linear growth with period 6528 has a surprisingly simple recipe:

Code: Select all

x = 22, y = 11, rule = B3/S23
2bo16bo$2ob2o13bobo$2bo14bo3bo$18bobo$19bo2$9b2o$9bobo$11bo8b2o$9bobo
8b2o$9b2o!

Haul: https://catagolue.appspot.com/haul/b3s2 ... 2074c774e2 (Rob Liston, 2020-01-22)

Easy 7G:

Code: Select all

x = 26, y = 19, rule = B3/S23
2bo$obo$b2o2$6bo$6bobo14bo$6b2o14bo$22b3o4$24bo$5bo17b2o$6bo16bobo$4b
3o$8b3o$10bo10b2o$9bo12b2o$21bo!

What happens if it is submitted to catagolue synthesis?
EDIT: how do I check if a synthesis like this is valid? (xp2_0o4cjiacggz116a9p643 in 24G)

Code: Select all

x = 129, y = 60, rule = B3/S23
25bo$26b2o$25b2o9$21bo75bobo$19b2o77b2o$20b2o76bo2$33bo66b3o$31b2o7bo$
17bo14b2o6bobo85bo$18bo21b2o84b2o$16b3o28bo79b2o$32b3o11b2o76bo$32bo
13bobo75bo$33bo90bo3$20bo36bobo50bo$15bobo2bobo34b2o49b3o$16b2o2b2o36b
o48bo3b2o$16bo90b3obobo$105b2o4bobo$106bobo4b2o$43bo62bobob3o$bo36b2o
2b2o63b2o3bo$b2o34bobo2bobo64b3o$obo36bo69bo3$26bo68bo$11bobo13bo67bo$
12b2o11b3o67bo$12bo28b3o47b2o$18b2o21bo50b2o$17bobo6b2o14bo48bo$19bo7b
2o$26bo90b3o2$38b2o81bo$39b2o79b2o$38bo81bobo9$33b2o$32b2o$34bo!

EDIT2: Help!

Code: Select all

x = 142, y = 142, rule = B3/S23
35bo$33bobo$34b2o13$44bo$42bobo$43b2o16$140bo$139bo$139b3o7$71b2o52bo$
71b2o51bo$124b3o2$72b2o$71bo2bo$71bo2bo$72b2o6$59b2o$59b2o$69bo$64b2o
2bobo$63bobob2ob2o13b2o$64bo20b2o3$82bo$81bobo$82b2o2$47b2o33bo$46bo2b
o32b2o$42b2o2bo2bo34bo$42b2o3b2o10bo22b2o$58b2o22bo10b2o3b2o$57bo34bo
2bo2b2o$58b2o32bo2bo$59bo33b2o2$58b2o$58bobo$59bo3$55b2o20bo$55b2o13b
2ob2obobo$71bobo2b2o$72bo$81b2o$81b2o6$68b2o$67bo2bo$67bo2bo$68b2o2$
15b3o$17bo51b2o$16bo52b2o7$3o$2bo$bo16$97b2o$97bobo$97bo13$106b2o$106b
obo$106bo!

[hkoenig]

how do I check if a synthesis like this is valid? (xp2_0o4cjiacggz116a9p643 in 24G)

Back up all the Gliders until they are coming from infinity. I tried that and this appears to be valid.

[Ian07]

What happens if it is submitted to catagolue synthesis?

Catagolue will most likely just reject it as invalid, since it's not designed to accept linear growth syntheses. The ones that are there (example) were added to the database manually by calcyman.

[GUYTU6J]

I submitted a 10G xs27_2llmzpie0fgkc synthesis minutes ago:

Code: Select all

x = 107, y = 33, rule = B3/S23
102bo$102bobo$102b2o2$46bo54bo$47bo51bobo$obo42b3o52b2o$b2o$bo$12bo$
13b2o50b2o38b2o$12b2o43b2o6b2o25b2o3b2o6b2o$57b2o33b2o3b2o2$13b2o33b2o
13b2o38b2o$12bobo32bobo13b2o38b2o$14bo34bo5$18b3o$18bo$19bo$97b2o$96bo
2bo$97b2o$60b2o$60bobo$60bo$53b2o$52bobo$54bo!

It suggests the following 8G component making that 4blocks+pi combo:

Code: Select all

x = 75, y = 52, rule = B3/S23
72bo$72bobo$72b2o2$21bo$bo17bobo$2bo17b2o$3o3$13bobo51bo$14b2o51bobo$
14bo52b2o11$41bo$42b2o$41b2o11$49bo$43bobo2bo$44b2o2b3o$44bo10$42b2o$
41bo2bo$42b2o!

Don't know how useful this kind of converter is, though.
EDIT: one of the blocks is unnecessary

Code: Select all

x = 107, y = 33, rule = B3/S23
80bobo19bo$81b2o19bobo$81bo20b2o2$101bo$99bobo$obo97b2o$b2o$bo$12bo$
13b2o50b2o38b2o$12b2o43b2o6b2o30b2o6b2o$57b2o38b2o2$13b2o48b2o38b2o$
12bobo48b2o38b2o$14bo5$18b3o$18bo$19bo$97b2o$96bo2bo$97b2o$60b2o$60bob
o$60bo$53b2o$52bobo$54bo!

But the component is still 8G:

Code: Select all

x = 75, y = 52, rule = B3/S23
obo69bo$b2o69bobo$bo70b2o2$21bo$19bobo$20b2o2$16bo$14bobo$15b2o50bo$
66bo$66b3o11$41bo$42b2o$41b2o11$49bo$43bobo2bo$44b2o2b3o$44bo10$42b2o$
41bo2bo$42b2o!

I've accidentally uploaded this failed synthesis to catagolue; what'll happen?

Code: Select all

$72bo$72bobo$72b2o2$21bo$bo17bobo$2bo17b2o$3o3$13bobo51bo$14b2o51bobo$
14bo52b2o11$41bo$42b2o$41b2o11$49bo$43bobo2bo$44b2o2b3o$44bo10$42b2o$
41bo2bo$42bobo$43bo!

[praosylen]

I submitted a 10G xs27_2llmzpie0fgkc synthesis minutes ago:
But the component is still 8G:

Code: Select all

rle

3G reduction to 7G/5G:

Code: Select all

x = 21, y = 26, rule = B3/S23
19bo$17b2o$18b2o7$6bobo$7b2o$7bo6bo$13bo$bo11b3o$obo14b2o$obo14bobo$b
o15bo7$19bo$18b2o$18bobo!

The pi/3-block combo turns out to make a block and a switch engine (and this block can also be replaced with a glider, making it a SE+2G collision).

[GUYTU6J]

What's wrong with this 27G xp3_o4im4id1qcz0dm0di9qi96zx11 synthesis? It cannot be uploaded to catagolue.

Code: Select all

x = 225, y = 37, rule = B3/S23
208bo$12bo194bo$12bobo192b3o$12b2o87bo101b3o$102bo$100b3o$o104bo$b2o
102bobo50bo$2o103b2o47b2o2bobo$154b2o2b2o3$119bo$43bo75bobo85b2o$6bo
34bobo8b2o5b2o41b2o5b2o8b2o25bo5b2o5b2o42b2obo2bo$4b2o36b2o3bo4bobo3bo
bo41bobo3bobo36bo4bobo3bobo29bo11bob2obo2bo10bo$5b2o41b2o3b2o3b2o43b2o
3b2o35b3o5b2o3b2o5b2o21bobo10bo5bob2o10bo$47b2o46bo20b3o46b2o22b2o10bo
b2obo2bo11bo$96bo19bo85bo2bobo2bo$4b2o88b3o20bo28b2o43bo11bo2bob2obo
10b2o$4bobo10b3o83b2o3b2o36b2o5b2o3b2o5b3o23bo10b2obo5bo10bobo$4bo12bo
84bobo3bobo41bobo3bobo4bo25bo10bo2bob2obo11bo$18bo38bo8b2o24b2o8b2o5b
2o41b2o5b2o5bo36bo2bob2o$57b2o6b2o24bobo110b2o$56bobo8bo25bo3$153b2o2b
2o$54b3o49b2o44bobo2b2o$56bo48bobo46bo$48bo6bo51bo$48b2o60b3o$47bobo
60bo$111bo95b3o$203b3o$205bo$204bo!

By the way, half-blockade + moose antler → ?-very_long_eater_siamese_eater_and_eater_and_block

Code: Select all

x = 67, y = 91, rule = B3/S23
65bo$64bo$64b3o2$2bo$obo$b2o25$15bobo$16b2o$16bo10$18b2o4bo$17bobo3b2o
$19bo3bobo43$62b3o$62bo$63bo!

The pi/3-block combo turns out to make a block and a switch engine (and this block can also be replaced with a glider, making it a SE+2G collision).

Ah! I didn't recognize that shape. I intended to make the component come entirely from above (in your recipe it is right).

[dvgrn]

What's wrong with this 27G xp3_o4im4id1qcz0dm0di9qi96zx11 synthesis? It cannot be uploaded to catagolue...

Looks like a valid synthesis, and it seems to be in there now.

By the way, half-blockade + moose antler → ?-very_long_eater_siamese_eater_and_eater_and_block

Code: Select all

x = 67, y = 91, rule = B3/S23
65bo$64bo$64b3o2$2bo$obo$b2o25$15bobo$16b2o$16bo10$18b2o4bo$17bobo3b2o
$19bo3bobo43$62b3o$62bo$63bo!

@Someone-who-knows-components, is that a previously known reaction? It's hard to know where to look for it in components_v4. It's the current recipe for xs24_31e8w8ozc8711dd, so I'm guessing it just got there and is therefore better than whatever was there before if anything.

(I sure wish Catagolue/Shinjuku had an audit trail so that "whatever was there before" is visible. Sometimes that's useful information, which is why it's so helpful that Mark Niemiec's database keeps multiple recipes.) (While I'm wishing, naturally the audit trail should include who uploaded what when.)

This half-blockade trick isn't exactly a component, I guess, since it only works to add it to an active pattern before the point of the hat shape appears. It could be a component if there's a way to get the closer block into place faster, in the form of a preblock instead of a pre-preblock at the moment when it first presents a two-cell surface.

That said, the trick works any time there's any way to sneak those two blocks in somehow. Which seems like it will be a good fraction of the time:

Code: Select all

#C modifications of xs23_ml1e8gzdl871
#C ( http://catagolue.appspot.com/object?apgcode=xs23_ml1e8gzdl871&rule=b3s23 )
#C Left, xs32_660u1jz69d1cd1da4; right, xs32_ml1e8w8ozdl8711dd.
x = 181, y = 99, rule = LifeSuper
173.O.O$95.M77.2O$93.M.M78.O$94.2M13$16.M.M$17.2M$17.M$77.O.O$77.2O4.
O.O$78.O4.2O$2.M81.O$M.M$.2M$131.M$129.M.M$130.2M4$142.2T$142.2T3$
141.2T$141.2T2$126.2M$125.M.M10.3M$127.M10.M$139.M$114.2M$115.2M$114.
M2$38.M73.M$36.M.M73.2M$37.2M72.M.M$48.2T$48.2T3$49.2T$49.2T5$33.2M$
32.M.M10.3M$34.M10.M$46.M$21.2M154.O$22.2M152.2O$21.M154.O.O2$19.M$
19.2M153.2O$18.M.M152.2O$175.O2$179.2O$178.2O$180.O9$81.O$80.2O$80.O.
O9$141.3O$141.O2.O$141.O$141.O$142.O!

Two-state:

Code: Select all

x = 256, y = 99, rule = B3/S23
248bobo$170bo77b2o$168bobo78bo$169b2o13$16bobo$17b2o$17bo$77bobo$77b2o
4bobo$78bo4b2o$2bo81bo$obo$b2o$206bo$204bobo$205b2o11$201b2o$200bobo
10b3o$202bo10bo$214bo$189b2o$190b2o$189bo2$38bo148bo$36bobo148b2o$37b
2o147bobo11$33b2o$32bobo10b3o$34bo10bo$46bo$21b2o229bo$22b2o227b2o$21b
o229bobo2$19bo$19b2o228b2o$18bobo227b2o$250bo2$254b2o$253b2o$255bo9$
81bo$80b2o$80bobo9$216b3o$216bo2bo$216bo$216bo$217bobo!

[apg]

(I sure wish Catagolue/Shinjuku had an audit trail so that "whatever was there before" is visible. Sometimes that's useful information, which is why it's so helpful that Mark Niemiec's database keeps multiple recipes.) (While I'm wishing, naturally the audit trail should include who uploaded what when.)

Shinjuku does keep all of the recipes in the graph (whereas Catagolue does indeed only keep the cheapest synthesis from that graph), and the git commit history is how you go 'back in time'.

[GUYTU6J]

@Someone-who-knows-components, is that a previously known reaction? It's hard to know where to look for it in components_v4. It's the current recipe for xs24_31e8w8ozc8711dd, so I'm guessing it just got there and is therefore better than whatever was there before if anything.

I observed that the still life got more instances in 4Glider_stdin than 5Glider_stdin, and it wasn't synthesized before my 6G recipe, so probably it is unknown. It arises from a LoM+2G explosion: (invalid example)

Code: Select all

x = 27, y = 14, rule = B3/S23
2bo$obo$b2o3$10bo$11b2o$10b2o13bo$23b2o$24b2o2$24bobo$24b2o$25bo!

16-cell predecessor:

Code: Select all

x = 12, y = 5, rule = B3/S23
o2bo$2bo$2bob3o2b2o$4bo3bo2bo$5b2o2b2o!

This half-blockade trick isn't exactly a component, I guess, since it only works to add it to an active pattern before the point of the hat shape appears. It could be a component if there's a way to get the closer block into place faster, in the form of a preblock instead of a pre-preblock at the moment when it first presents a two-cell surface.

It feels like an active species formed in situ in an organic reaction. A half-depressed LoM reaction won't work if the tip is present due to B3q - see gen 45 and 55:

Code: Select all

x = 20, y = 43, rule = LifeHistory
2A5.2A$A7.A$.3A.3A$3.A.A$4.D2$2.2D.2D8$2.2A$3.2A$2.A4.2A$7.A.A$7.A22$
18.2A$17.2A$19.A!

Anyway, good job for the modification on the newly-synthesized still life. (In 4 glider syntheses thread)

[dvgrn]

Shinjuku does keep all of the recipes in the graph (whereas Catagolue does indeed only keep the cheapest synthesis from that graph), and the git commit history is how you go 'back in time'.

In terms of accessibility to a casual researcher, that seems to be up there with

“But the plans were on display…”
“On display? I eventually had to go down to the cellar to find them.”
“That’s the display department.”
“With a flashlight.”
“Ah, well, the lights had probably gone.”
“So had the stairs.”
“But look, you found the notice, didn’t you?”
“Yes,” said Arthur, “yes I did. It was on display in the bottom of a locked filing cabinet stuck in a disused lavatory with a sign on the door saying ‘Beware of the Leopard.”

Maybe a script that could be written to improve the convenience of retrieval a little bit. Or maybe the leopard isn't as scary as I think it is.

I'm just a little bit wary of the Shinjuku leopard, because it ate several hours of my time the last time I tried wrestling with it, and I was never even able to read the documentation, let alone do anything useful. I think I do have everything in place now, now that Freywa gave away the secret about moving the Juypiter notebook files up one level -- just haven't had the energy to tackle trying to learn the system again.

Anyway, the current state of Shinjuku recipe-retrieval art is definitely not quite on a par yet with the downloadable RLE stamp collections of past syntheses on Mark Niemiec's website.