Golly 3.0b2

For general discussion about Conway's Game of Life.
User avatar
Andrew
Moderator
Posts: 919
Joined: June 2nd, 2009, 2:08 am
Location: Melbourne, Australia
Contact:

Golly 3.0b2

Post by Andrew » August 2nd, 2017, 9:26 pm

EDIT: Golly 3.0b3 has been released so the links to 3.0b2 have been removed.

This version has major improvements to the Larger than Life algorithm:

- Support for unbounded universes, within certain limits. See Help > Algorithms > Larger than Life, which also has a new History section and more references, thanks to Kellie Evans.

- Patterns run much faster, thanks to some fast algorithms from Adam Goucher (aka calcyman) and Dean Hickerson, and further optimizations by myself and Chris Rowett.

- The maximum range is now 500. See Patterns/Larger-than-Life/ for a couple of spaceships in range-100 rules, and a spaceship in a range-500 rule (created by scaling up the R100 spaceship).

If all goes well we're planning to do the final 3.0 release in late August.
Use Glu to explore CA rules on non-periodic tilings: DominoLife and HatLife

User avatar
muzik
Posts: 5614
Joined: January 28th, 2016, 2:47 pm
Location: Scotland

Re: Golly 3.0b2

Post by muzik » August 2nd, 2017, 9:29 pm

This is exciting.

That long-awaited 50c ship:

Code: Select all

x = 100, y = 100, rule = R99,C3,M0,S0..0,B5000..5000,NM
100A$100A$100A$100A$100A$100A$100A$100A$100A$100A$100A$100A$100A$100A
$100A$100A$100A$100A$100A$100A$100A$100A$100A$100A$100A$100A$100A$
100A$100A$100A$100A$100A$100A$100A$100A$100A$100A$100A$100A$100A$100A
$100A$100A$100A$100A$100A$100A$100A$100A$100A$100B$100B$100B$100B$
100B$100B$100B$100B$100B$100B$100B$100B$100B$100B$100B$100B$100B$100B
$100B$100B$100B$100B$100B$100B$100B$100B$100B$100B$100B$100B$100B$
100B$100B$100B$100B$100B$100B$100B$100B$100B$100B$100B$100B$100B$100B
$100B$100B$100B$100B$100B!
Since 500c is now obviously our upper speed limit (since i doubt it'll be pushed up beyond that, bu one can dream), the challenge to find ships and replicators travelling that fast begins.
Last edited by muzik on August 2nd, 2017, 9:49 pm, edited 1 time in total.

User avatar
Saka
Posts: 3627
Joined: June 19th, 2015, 8:50 pm
Location: Indonesia
Contact:

Re: Golly 3.0b2

Post by Saka » August 2nd, 2017, 9:41 pm

Aaaaaaaaaaaaaaaaaaaaaaaaaaaamazing!!!!!!
When can we start making 500c/1 ships???

User avatar
muzik
Posts: 5614
Joined: January 28th, 2016, 2:47 pm
Location: Scotland

Re: Golly 3.0b2

Post by muzik » August 2nd, 2017, 9:58 pm

250c in a quarter of a million cells:

Code: Select all

x = 500, y = 500, rule = R499,C3,M0,S0..0,B125000..125000,NM
500A$500A$500A$500A$500A$500A$500A$500A$500A$500A$500A$500A$500A$500A
$500A$500A$500A$500A$500A$500A$500A$500A$500A$500A$500A$500A$500A$
500A$500A$500A$500A$500A$500A$500A$500A$500A$500A$500A$500A$500A$500A
$500A$500A$500A$500A$500A$500A$500A$500A$500A$500A$500A$500A$500A$
500A$500A$500A$500A$500A$500A$500A$500A$500A$500A$500A$500A$500A$500A
$500A$500A$500A$500A$500A$500A$500A$500A$500A$500A$500A$500A$500A$
500A$500A$500A$500A$500A$500A$500A$500A$500A$500A$500A$500A$500A$500A
$500A$500A$500A$500A$500A$500A$500A$500A$500A$500A$500A$500A$500A$
500A$500A$500A$500A$500A$500A$500A$500A$500A$500A$500A$500A$500A$500A
$500A$500A$500A$500A$500A$500A$500A$500A$500A$500A$500A$500A$500A$
500A$500A$500A$500A$500A$500A$500A$500A$500A$500A$500A$500A$500A$500A
$500A$500A$500A$500A$500A$500A$500A$500A$500A$500A$500A$500A$500A$
500A$500A$500A$500A$500A$500A$500A$500A$500A$500A$500A$500A$500A$500A
$500A$500A$500A$500A$500A$500A$500A$500A$500A$500A$500A$500A$500A$
500A$500A$500A$500A$500A$500A$500A$500A$500A$500A$500A$500A$500A$500A
$500A$500A$500A$500A$500A$500A$500A$500A$500A$500A$500A$500A$500A$
500A$500A$500A$500A$500A$500A$500A$500A$500A$500A$500A$500A$500A$500A
$500A$500A$500A$500A$500A$500A$500A$500A$500A$500A$500A$500A$500A$
500A$500A$500A$500A$500A$500A$500A$500B$500B$500B$500B$500B$500B$500B
$500B$500B$500B$500B$500B$500B$500B$500B$500B$500B$500B$500B$500B$
500B$500B$500B$500B$500B$500B$500B$500B$500B$500B$500B$500B$500B$500B
$500B$500B$500B$500B$500B$500B$500B$500B$500B$500B$500B$500B$500B$
500B$500B$500B$500B$500B$500B$500B$500B$500B$500B$500B$500B$500B$500B
$500B$500B$500B$500B$500B$500B$500B$500B$500B$500B$500B$500B$500B$
500B$500B$500B$500B$500B$500B$500B$500B$500B$500B$500B$500B$500B$500B
$500B$500B$500B$500B$500B$500B$500B$500B$500B$500B$500B$500B$500B$
500B$500B$500B$500B$500B$500B$500B$500B$500B$500B$500B$500B$500B$500B
$500B$500B$500B$500B$500B$500B$500B$500B$500B$500B$500B$500B$500B$
500B$500B$500B$500B$500B$500B$500B$500B$500B$500B$500B$500B$500B$500B
$500B$500B$500B$500B$500B$500B$500B$500B$500B$500B$500B$500B$500B$
500B$500B$500B$500B$500B$500B$500B$500B$500B$500B$500B$500B$500B$500B
$500B$500B$500B$500B$500B$500B$500B$500B$500B$500B$500B$500B$500B$
500B$500B$500B$500B$500B$500B$500B$500B$500B$500B$500B$500B$500B$500B
$500B$500B$500B$500B$500B$500B$500B$500B$500B$500B$500B$500B$500B$
500B$500B$500B$500B$500B$500B$500B$500B$500B$500B$500B$500B$500B$500B
$500B$500B$500B$500B$500B$500B$500B$500B$500B$500B$500B$500B$500B$
500B$500B$500B$500B$500B$500B$500B$500B$500B$500B$500B$500B$500B$500B
!
It works, I guess - and with absolutely no lag on 10^0!

User avatar
rowett
Moderator
Posts: 3776
Joined: January 31st, 2013, 2:34 am
Location: UK
Contact:

Re: Golly 3.0b2

Post by rowett » August 3rd, 2017, 5:55 am

Andrew wrote:The 2nd beta version of Golly 3.0 is now available
This build also supports MAP rules with Hex or von Neumann neighborhoods:

Hexagonal neighborhoods have 7 cells (center plus 6 neighbors) which gives 128 (2^7) possible combinations of cells. These encode into 22 base64 characters.
Von Neumann neighborhoods have 5 cells (center plus 4 neighbors) which gives 32 (2^5) possible combinations of cells. These encode into 6 base 64 characters.

User avatar
Saka
Posts: 3627
Joined: June 19th, 2015, 8:50 pm
Location: Indonesia
Contact:

Re: Golly 3.0b2

Post by Saka » August 3rd, 2017, 7:06 am

I'm still disappointed that we can't have APG-LTL yet because we can't make ruleTables :(
And time to edit all the files of my WIP collection...

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

Re: Golly 3.0b2

Post by dvgrn » August 3rd, 2017, 7:09 am

Andrew wrote:The 2nd beta version of Golly 3.0 is now available
Just ran into what looks like a leftover OpenGL problem on my old laptop, showing up on Larger than Life rules with a lot of states.
ModernArt looks different on different scales
ModernArt looks different on different scales
ModernArtScaleComparison.png (33.37 KiB) Viewed 12870 times
I've attached a few more screenshots to give a better sense of the problem. This is with the color gradient turned off in Preferences. With the gradient turned on, sometimes I get all-white cells for any state, or sometimes (on another run of Golly) I'll get a yellow or green color, with square or rectangular pixels shaded in odd ways. See ModernArtT32scale1-to-8.png in the ZIP attachment. Seems to be using the contents of some random memory buffer to draw the pixel.

I used to see this kind of thing for LifeHistory or other multistate rules. Now that algo seems to be fine, but Larger Than Life is still showing the same behavior.

The gradual fading in documented in the screenshots doesn't always happen. I just tried a run after looking at LifeHistory, and oddly enough the LifeHistory colors (green, blue, white, red, yellow, gray) were spread proportionally across all 255 ON states, replacing the yellow-to-red color gradient.
Attachments
ModernArtscreenshots.zip
screenshots showing cell states fading gradually into visibility as the state number gets higher
(418.09 KiB) Downloaded 280 times

drc
Posts: 1664
Joined: December 3rd, 2015, 4:11 pm

Re: Golly 3.0b2

Post by drc » August 3rd, 2017, 8:38 am

I'm having the same problem, in the attached picture.

Note that I upgraded my computer to support OpenGL again, but since dvgrn posted about having this same problem I believe it may be a separate issue.
Attachments
A 255-state LtL rule on the highest scale possible
A 255-state LtL rule on the highest scale possible
huh.png (60.09 KiB) Viewed 12860 times

User avatar
Andrew
Moderator
Posts: 919
Joined: June 2nd, 2009, 2:08 am
Location: Melbourne, Australia
Contact:

Re: Golly 3.0b2

Post by Andrew » August 3rd, 2017, 9:39 am

dvgrn wrote:Just ran into what looks like a leftover OpenGL problem on my old laptop, showing up on Larger than Life rules with a lot of states.
I'm a bit confused. Are you saying that LtL rendering was fine in 3.0b1 but not in 3.0b2? That seems very odd given that the rendering code barely changed (from memory -- I'll have to double check).

Some more questions:

Do you see any rendering problems at scale 1:1?

From the screenshots it seems that rendering is ok at scale 1:4 and the problems only start at 1:8 and beyond?

If you remove the suffix (via Set Rule) from the rule used in ModernArt.mcl, does that fix the rendering problems?

Do the rendering problems only occur when the number of states is > 2?
Use Glu to explore CA rules on non-periodic tilings: DominoLife and HatLife

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

Re: Golly 3.0b2

Post by dvgrn » August 3rd, 2017, 9:47 am

Andrew wrote:
dvgrn wrote:Just ran into what looks like a leftover OpenGL problem on my old laptop, showing up on Larger than Life rules with a lot of states.
I'm a bit confused. Are you saying that LtL rendering was fine in 3.0b1 but not in 3.0b2? That seems very odd given that the rendering code barely changed (from memory -- I'll have to double check).
Can't tell you for sure -- I've tested Larger Than Life on several systems, and it's quite possible that I didn't look at ModernArt.mcl with 3.0b1.
Andrew wrote:Do you see any rendering problems at scale 1:1?
No.
Andrew wrote:From the screenshots it seems that rendering is ok at scale 1:4 and the problems only start at 1:8 and beyond?
Correct. No visible problems at scale 1:4, 1:2, etc.
Andrew wrote:Do the rendering problems only occur when the number of states is > 2?
I didn't see any trouble with the sample two-state patterns I looked at.
Andrew wrote:If you remove the suffix (via Set Rule) from the rule used in ModernArt.mcl, does that fix the rendering problems?
I won't be back to my old laptop until tonight -- will check then.

User avatar
rowett
Moderator
Posts: 3776
Joined: January 31st, 2013, 2:34 am
Location: UK
Contact:

Re: Golly 3.0b2

Post by rowett » August 3rd, 2017, 2:56 pm

I have a working theory for the rendering issue. Please let me know if you are seeing it without one of the conditions below:
  • OpenGL 1.x
  • multi state rule where zoom x states > maximum texture size
Example: 40 state rule at 32x zoom with maximum texture size of 1024 would render badly but would work at 16x zoom and lower.

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

Re: Golly 3.0b2

Post by dvgrn » August 3rd, 2017, 10:55 pm

Andrew wrote:If you remove the suffix (via Set Rule) from the rule used in ModernArt.mcl, does that fix the rendering problems?
No difference with bounded vs. unbounded universes.
rowett wrote:I have a working theory for the rendering issue. Please let me know if you are seeing it without one of the conditions below:
  • OpenGL 1.x
  • multi state rule where zoom x states > maximum texture size
Example: 40 state rule at 32x zoom with maximum texture size of 1024 would render badly but would work at 16x zoom and lower.
This seems to be almost exactly right. I found that changing the ModernArt rule to 33 states or less seemed to fix all the display glitches. The glitches get to be kind of intermittent as the number of states goes below 40 -- sometimes everything looks good at first, then gets weird if you zoom in and out a lot. But they never quite go away at 1:32 scale until there are less than 34 states.

The multiplication doesn't quite work out -- the transition ought to be at 32, not 33 -- but it sure looks like you're on the right track. Changing from 255 states to 127 made the problem disappear at 1:8 scale but it was still there at 1:16, for example.

If I modify your theory to exclude state 0 -- which might make sense, since state 0 never has an icon --
  • multi state rule where zoom x nonzero states > maximum texture size
then I think it matches my observations perfectly... so far, anyway.

User avatar
muzik
Posts: 5614
Joined: January 28th, 2016, 2:47 pm
Location: Scotland

Re: Golly 3.0b2

Post by muzik » August 4th, 2017, 8:16 am

The example patterns for Larger than Life are still on a torus. Is this intentional?

User avatar
rowett
Moderator
Posts: 3776
Joined: January 31st, 2013, 2:34 am
Location: UK
Contact:

Re: Golly 3.0b2

Post by rowett » August 4th, 2017, 8:55 am

dvgrn wrote:Just ran into what looks like a leftover OpenGL problem on my old laptop, showing up on Larger than Life rules with a lot of states.
I just committed a potential fix. Please let me know if it helps.

User avatar
BlinkerSpawn
Posts: 1992
Joined: November 8th, 2014, 8:48 pm
Location: Getting a snacker from R-Bee's

Re: Golly 3.0b2

Post by BlinkerSpawn » August 4th, 2017, 8:58 am

muzik wrote:The example patterns for Larger than Life are still on a torus. Is this intentional?
You can edit the suffixes out manually under "Set Rule" and re-save them if it bothers you.
LifeWiki: Like Wikipedia but with more spaceships. [citation needed]

Image

User avatar
muzik
Posts: 5614
Joined: January 28th, 2016, 2:47 pm
Location: Scotland

Re: Golly 3.0b2

Post by muzik » August 4th, 2017, 9:01 am

BlinkerSpawn wrote:
muzik wrote:The example patterns for Larger than Life are still on a torus. Is this intentional?
You can edit the suffixes out manually under "Set Rule" and re-save them if it bothers you.
That I know, it just feels like more of a remnant/thing missed out to me.

The newer additions to the folder don't have this limit either.

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

Re: Golly 3.0b2

Post by dvgrn » August 4th, 2017, 11:34 am

rowett wrote:
dvgrn wrote:Just ran into what looks like a leftover OpenGL problem on my old laptop, showing up on Larger than Life rules with a lot of states.
I just committed a potential fix. Please let me know if it helps.
Unfortunately I won't be back to that laptop again for about a week now -- I'm traveling to upstate New York, where my rather rustic accommodations will not include amenities like electricity, let alone Internet access or cell phone service.

@drc, might you be able to test this if someone builds a new executable?

User avatar
muzik
Posts: 5614
Joined: January 28th, 2016, 2:47 pm
Location: Scotland

Re: Golly 3.0b2

Post by muzik » August 4th, 2017, 1:51 pm

I know the ability for Golly to simulate non-totalistic rules is old news, but can these be added to the respective folder? They seem to qualify as interesting enough:

Code: Select all

# Rule containing a set of small, extensible, adjustable-speed spaceships
# for any integer c/2n, n>1. Found by AbhpzTa.
# For more patterns in this rule see the thread:
# http://www.conwaylife.com/forums/viewtopic.php?f=11&t=3008
x = 20, y = 15, rule = B2c3ae4ai56c_S2-kn3-enq4
3bo5bo6bo2bo$2bob6o5bob3o$3bo5bo6bo2bo4$2bo6bo5bo3bo$bob7o4bob4o$2bo6b
o5bo3bo4$bo7bo4bo4bo$ob8o3bob5o$bo7bo4bo4bo!

Code: Select all

# Rule containing a set of small, extensible, adjustable-speed spaceships
# for any integer c/2n+1, n>1. Found by Aidan F. Pierce.
# For more patterns in this rule see the thread:
# http://www.conwaylife.com/forums/viewtopic.php?f=11&t=3007
x = 20, y = 17, rule = B2c3aj4nrt5i6c78_S1c23enr4aet5-iq67
2bo4b2o6bob2o$2b5o2bo5b2o2bo$2bo4b2o6bob2o5$bo5b2o5bo2b2o$b6o2bo4b3o2b
o$bo5b2o5bo2b2o5$o6b2o4bo3b2o$7o2bo3b4o2bo$o6b2o4bo3b2o!

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

Re: Golly 3.0b2

Post by dvgrn » August 4th, 2017, 2:10 pm

muzik wrote:I know the ability for Golly to simulate non-totalistic rules is old news, but can these be added to the respective folder? They seem to qualify as interesting enough...
What do you think of the unit-fraction spaceship stunt-flying pattern in the latest Golly beta?

Maybe that orthogonal-spaceships pattern could be updated to fill in the odd fractions with B2c3aj4nrt5i6c78_S1c23enr4aet5-iq67 spaceships, and then the comments from these two patterns could be added to the pattern info. I guess that would require a RainbowASOv1.4, though...?

User avatar
Andrew
Moderator
Posts: 919
Joined: June 2nd, 2009, 2:08 am
Location: Melbourne, Australia
Contact:

Re: Golly 3.0b2

Post by Andrew » August 5th, 2017, 7:01 am

drc wrote:I'm having the same problem, ...
@drc: Please try this version (3.0b3) and let us know if it solves all your rendering problems:

http://www.trevorrow.com/golly/golly-3. ... -64bit.zip

This is a Windows 64-bit build. Let me know if you need a 32-bit build.
Use Glu to explore CA rules on non-periodic tilings: DominoLife and HatLife

drc
Posts: 1664
Joined: December 3rd, 2015, 4:11 pm

Re: Golly 3.0b2

Post by drc » August 5th, 2017, 8:20 am

I can't just yet because my computer is broken, but I'll try asap

User avatar
muzik
Posts: 5614
Joined: January 28th, 2016, 2:47 pm
Location: Scotland

Re: Golly 3.0b2

Post by muzik » August 5th, 2017, 8:26 pm

dvgrn wrote:
muzik wrote:I know the ability for Golly to simulate non-totalistic rules is old news, but can these be added to the respective folder? They seem to qualify as interesting enough...
What do you think of the unit-fraction spaceship stunt-flying pattern in the latest Golly beta?

Maybe that orthogonal-spaceships pattern could be updated to fill in the odd fractions with B2c3aj4nrt5i6c78_S1c23enr4aet5-iq67 spaceships, and then the comments from these two patterns could be added to the pattern info. I guess that would require a RainbowASOv1.4, though...?
Well I am working on a version that contains 2c/n spaceships as well (although the 2c/even ships are cheatishly filled in with perfect ships instead), so that rule could also be used. Only problem is that it removes all the odd perfect spaceships and replaces them with the newer adjustable odd speeds rule, but the colours could still be there if the 2c/n ships can be worked into it.

I don't see why the two adjustable rules shouldn't get a mention on their own, though - they stand out pretty well from most other non-totalistic rules.

User avatar
gameoflifeboy
Posts: 474
Joined: January 15th, 2015, 2:08 am

Re: Golly 3.0b2

Post by gameoflifeboy » August 8th, 2017, 10:03 pm

I just noticed that the survival (and presumably the same thing happens with the birth) conditions of Larger than Life rules are converted to the equivalent values modulo 2^32 between -2^31 and 2^31 - 1. This has some interesting consquences, such as how R1,C0,M0,S-4294967294..-4294967293,B3..3,NM is equivalent to Life. It also means that R1,C0,M0,S0..<some large number>,B3..3,NM is only valid if the large number is less than 2^31 mod 2^32.

Interestingly, R1,C0,M0,S0..2147483647,B3..3,NM still works, and doesn't auto-correct to R1,C0,M0,S0..8,B3..3,NM like it should.

Also, the default directory where custom rules get added is the main golly-3.0-b2 directory, instead of ~/.golly/rules like it should be.

User avatar
Andrew
Moderator
Posts: 919
Joined: June 2nd, 2009, 2:08 am
Location: Melbourne, Australia
Contact:

Re: Golly 3.0b2

Post by Andrew » August 9th, 2017, 8:19 pm

gameoflifeboy wrote:I just noticed that the survival (and presumably the same thing happens with the birth) conditions of Larger than Life rules are converted to the equivalent values modulo 2^32 between -2^31 and 2^31 - 1. ...
Yep, that's very confusing. The final 3.0 release will check that the S and B limits are from 0 to the neighborhood size: (2R+1)^2 for NM rules, or 2R(R+1)+1 for NN rules.
Also, the default directory where custom rules get added is the main golly-3.0-b2 directory, instead of ~/.golly/rules like it should be.
I don't see any such problem on my Ubuntu 12 system so it must be yet another glitch in your version of Ubuntu. You should be able to set the path using Preferences > Control. If that's not possible due to the dialog bug on your Ubuntu then quit Golly and edit the user_rules path in GollyPrefs.
Use Glu to explore CA rules on non-periodic tilings: DominoLife and HatLife

AforAmpere
Posts: 1334
Joined: July 1st, 2016, 3:58 pm

Re: Golly 3.0b2

Post by AforAmpere » August 9th, 2017, 9:33 pm

I know I've already asked this, but algorithms have changed since then. Could B and S conditions be added like B3..4,6..8,10, S2..6,7..9,11,13? This format would allow previously defined rules to still work, but allow more. This addition would allow more exploration of the LTL rulespace, and it would allow better tweaking of rules, so that interesting rules can be improved. I obviously am not asking for this anytime soon, but if it could be implemented in the future.
I manage the 5S project, which collects all known spaceship speeds in Isotropic Non-totalistic rules. I also wrote EPE, a tool for searching in the INT rulespace.

Things to work on:
- Find (7,1)c/8 and 9c/10 ships in non-B0 INT.
- EPE improvements.

Post Reply