Page 11 of 11

Re: 16 in 16: Efficient 16-bit Synthesis Project

PostPosted: May 23rd, 2017, 1:26 am
by yootaa
16.995 in 8G:
x = 50, y = 35, rule = B3/S23
18bo$19bo$17b3o5$29bo$27b2o$28b2o$19bo$19b2o$18bobo$36bobo$36b2o$37bo
4$b2o$obo$2bo$34b2o$33b2o$35bo$29b2o$30b2o$29bo5$47b3o$47bo$48bo!


EDIT:
Jormungant wrote:I found an approach that may work for 16.1787 (aka xs16_069m4koz311); however, I don't know if there is a 3 or 4 glider construction for a transient group that has 7 cells at some point (found on the upper right).

x = 21, y = 23, rule = B3/S23
13bo$11b2o$12b2o$7bo$7b2o9b3o$bo4bobo8bo$2b2o13bo2bo$b2o14bo2$10bo$9b
2o$9bobo$5b2o$4bobo$6bo4$b2o$obo$2bo11bo$14b2o$13bobo!

In 11G:
x = 46, y = 43, rule = B3/S23
31bo$30bo$30b3o3$bo$2bo$3o30bobo$33b2o$34bo5$33bobo$33b2o$29b2o3bo$28b
2o$30bo5$6b2o$5bobo$7bo$44b2o$43b2o$45bo$28b2o$28bobo$28bo$4b2o$5b2o$
4bo4$2o$b2o$o12b2o$12bobo$14bo!

Re: 16 in 16: Efficient 16-bit Synthesis Project

PostPosted: May 23rd, 2017, 4:41 am
by Goldtiger997
Jormungant wrote:I found an approach that may work for 16.1787...


16.1787 in 10 gliders:

x = 26, y = 30, rule = B3/S23
18bo$18bobo$18b2o3$2bo$obo$b2o2$24bo$6b3o14b2o$8bo10bo3bobo$7bo12b2o$
19b2o2$23b3o$15b3o5bo$15bo8bo$5bo10bo$5b2o$4bobo4$bo$b2o$obo$13b3o$15b
o$14bo!


EDIT: missed yootaa's post, but this is cheaper anyway.

EDIT2:

16.810 in 10 gliders:

x = 37, y = 36, rule = B3/S23
22bobo10bo$23b2o8b2o$23bo10b2o3$24bobo$24b2o$14bobo8bo$14b2o$15bo$8bob
o$8b2o$9bo5b3o$15bo$o3bo11bo$b2obobo$2o2b2o$17b2o$17bobo$17bo14$35b2o$
34b2o$36bo!

Re: 16 in 16: Efficient 16-bit Synthesis Project

PostPosted: May 23rd, 2017, 9:26 am
by Jormungant
16.1929 (aka xs16_0g5r8b5z121) in 10 gliders.

x = 62, y = 58, rule = B3/S23
o$b2o$2o$54bo$54bobo$54b2o$21bobo$22b2o$22bo26bo$48bo$48b3o11$28bo$27b
o$27b3o$24bo$25bo$23b3o5$31bobo$32b2o$32bo2$33b2o$33bobo$33bo2$28b2o$
29b2o$28bo14$60b2o$59b2o$61bo!


16.1717 (aka xs16_4aajk46zx121) in 11 gliders.

x = 47, y = 30, rule = B3/S23
25bo$23b2o$24b2o4$29bo$28bo$28b3o4$23b2o$23bobo$13bo9bo$11bobo$12b2o$
25b3o$25bo$26bo$13b2o$14b2o$13bo4b2o24b3o$2o15bobo24bo$b2o16bo25bo$o
21b2o$23b2o$22bo7b2o$30bobo$30bo!

Re: 16 in 16: Efficient 16-bit Synthesis Project

PostPosted: May 24th, 2017, 7:17 am
by chris_c
So just 16.836 to go, is that correct? It can be done by making a 1G optimisation to the existing 16G synthesis. There is a B and a loaf that is made in 5G. Instead it can be made in 4G by colliding a block with a 2G loaf+blinker. Frustratingly my phone line is down so I can't post the exact details or make a push to my github repo...

Re: 16 in 16: Efficient 16-bit Synthesis Project

PostPosted: May 24th, 2017, 9:07 am
by Goldtiger997
chris_c wrote:So just 16.836 to go... Frustratingly my phone line is down so I can't post the exact details or make a push to my github repo...


Bad lack about the phone line. I tried 32 soups, and the best I found is only slightly better.

16.836 in 13 gliders:

x = 89, y = 66, rule = B3/S23
74bo$74bobo$74b2o6$7bo$5bobo$6b2o11$26bo$24bobo$25b2o5$35bo$36bo$34b3o
3$44bo$42b2o$43b2o2$40bo$31b3o5bo$33bo5b3o$32bo$47b2o$46bobo$48bo$50b
2o$50bobo$36b2o12bo$35b2o$37bo6$48b3o$48bo$49bo2$2o$b2o$o3$86b2o$86bob
o$86bo!


Now all 16-bit still-lifes can be synthesised in less than 16 gliders!!!

Thanks to chris_c, BlinkerSpawn, AbptzTa, Extrementhusiast, Sokwe, BobShemyakin, yootaa, Jormagant, dvgrn, and Bullet51 for your contributions (sorry if I missed anyone)!

Re: 16 in 16: Efficient 16-bit Synthesis Project

PostPosted: May 24th, 2017, 9:24 am
by chris_c
Goldtiger997 wrote:16.836 in 13 gliders


Yes, last one down! I'm back online and pushed what will the last commit for a while. All 16-bit still lifes are synthesisable in at most 15G with an average of at most 10.47G!

Thanks to everyone!

Re: 16 in 16: Efficient 16-bit Synthesis Project

PostPosted: May 24th, 2017, 12:56 pm
by BobShemyakin
Great work! Thanks to everyone!

Bob Shemyakin

Re: 16 in 16: Efficient 16-bit Synthesis Project

PostPosted: May 25th, 2017, 6:59 am
by Gamedziner
Congratulations!

Re: 16 in 16: Efficient 16-bit Synthesis Project

PostPosted: May 27th, 2017, 3:01 pm
by dvgrn
Something I thought of recently: sometime prior to tackling the Still Lifes With A Prime Number of Bits That Must Not Be Named, wouldn’t it make sense to put together a helper utility script for finding likely 3G/4G/5G spark-making reactions?

That script of chris_c's for finding sparkish stuff by population has been very handy, but on multiple runs it ends up painfully re-generating the same population sequences. Why not just write out text files recording the population sequences for each pattern in each gencols output file, and then do a simple text search for the particular sequence that’s wanted?

Seems as if that wouldn’t take much more storage space than the original gencols output, and searches would be thousands of times faster. One byte per population value would be good enough -- no need to record population counts for the big messes that end up over 200 cells or so...?

Maybe it’s worth checking in to Github collision files that are as non-redundant as possible, and also as complete as possible, so that nothing else along the lines of the three-glider switch engine gets missed.

I don’t think it will be that hard to avoid redundancy altogether, with some careful work. Ultimately I’d like to combine this idea with the Enumerating Three-Glider Collisions project, so that for 3G collisions there are maybe 72 collision files checked in -- one for each of the 71 glider collisions with a third glider hitting the reaction at T>=1, and one for all the simultaneous 3G collisions.

Re: 16 in 16: Efficient 16-bit Synthesis Project

PostPosted: May 30th, 2017, 1:22 pm
by AbhpzTa
GitHub's list is not included this edit.
Please include it.

Re: 16 in 16: Efficient 16-bit Synthesis Project

PostPosted: May 30th, 2017, 1:50 pm
by chris_c
AbhpzTa wrote:GitHub's list is not included this edit.
Please include it.


OK, done. I also found this minor improvement to 16.712:

x = 41, y = 38, rule = B3/S23
34bo$34bobo$34b2o5$9bo$8bo$8b3o3$7b3o$7bo$8bo$22bo$22bo$22bo5$35b3o$
35bo$bo34bo$b2o$obo3$8b2o4b2o$9b2o4b2o21b2o$8bo5bo23bobo$38bo$11b2o$
10bobo$12bo20b3o$33bo$34bo!

Re: 16 in 16: Efficient 16-bit Synthesis Project

PostPosted: July 11th, 2017, 7:23 pm
by wwei23
Where's the 17 in 17 synthesis project?

Re: 16 in 16: Efficient 16-bit Synthesis Project

PostPosted: July 11th, 2017, 8:36 pm
by BlinkerSpawn
wwei23 wrote:Where's the 17 in 17 synthesis project?

As was said in another post somewhere, the current focus is getting a more functional, lower-maintenance system in place for depositing, archiving, and locating syntheses.
Once that's done, 17-bit syntheses will be easy to work with.
Using the infrastructure currently in place, though, it's a pretty tall order.

Re: 16 in 16: Efficient 16-bit Synthesis Project

PostPosted: July 13th, 2017, 11:53 am
by gmc_nxtman
Cross-posting from the construction practice thread, because I'm pretty sure this is an improvement (16.785):

x = 46, y = 51, rule = B3/S23
4bo$5b2o$4b2o5$26bo$11bo12bobo$9bobo13b2o$10b2o$43bo$43bobo$38bo4b2o$
38bobo$38b2o2$36bo$37bo$35b3o23$b2o$2b2o$bo2$43bo$42b2o$4b2o36bobo$3bo
bo$5bo!

Re: 16 in 16: Efficient 16-bit Synthesis Project

PostPosted: July 15th, 2017, 8:53 am
by Goldtiger997
AbhpzTa wrote:16.1391 in 7 gliders:
x = 23, y = 23, rule = B3/S23
9bo$9bobo$9b2o2$5bobo$6b2o$6bo2$21bo$20bo$8bobo9b3o$9b2o5b3o$9bo6bo$
17bo$9b2o$8b2o$10bo4$bo$b2o$obo!


I was trawling through the last pages of this thread, when I found a small reduction to the above synthesis.

16.1391 in 6 gliders:

x = 31, y = 23, rule = B3/S23
29bo$28bo$28b3o3$13bobo$14b2o$7b3o4bo$9bo6bo$8bo7b2o$15bobo$24b3o$24bo
$25bo7$bo$b2o$obo!


chris_c wrote:A preliminary step in this project was creating a translation between Niemiec's still life numbering and apgcodes. The fact that a similar list does not exist for oscillators is the only reason that I didn't reply to Goldtiger's query here.


After seeing Apple Bottom's great work in the wiki, with most small patterns having an auto-generated synthesis, but oscillators showing an invalid pattern message, I was reminded of the above quote. Couldn't a simple numbering system for oscillators be created by using apgcodes and ordering them alphabetically?

Re: 16 in 16: Efficient 16-bit Synthesis Project

PostPosted: July 15th, 2017, 2:24 pm
by wwei23
I've been filing away glider+object collisions, does anyone want to see?
Also, if a still life is symmetric along the red, then these two collisions are the same:
x = 35, y = 16, rule = LifeHistory
33.A$13.A18.A$12.A19.3A$12.3A3$9.D19.D$8.D19.D$7.D19.D$6.D19.D$5.D19.
D$4.D19.D$3.D19.D$2.D19.D$.D19.D$D19.D!

Because the second one in two generations looks like this:
x = 15, y = 15, rule = LifeHistory
12.A$12.A.A$12.2A3$9.D$8.D$7.D$6.D$5.D$4.D$3.D$2.D$.D$D!

Which reflected along the red looks like this:
x = 15, y = 15, rule = LifeHistory
13.A$12.A$12.3A3$9.D$8.D$7.D$6.D$5.D$4.D$3.D$2.D$.D$D!

Which is the same as the first collision.