Helix glider; VTI file; Smoothlife (was Re: glider in 3D U-Skate world)
20 views
Skip to first unread message
Robert Munafo
Oct 4, 2012, 9:58:53 AM10/4/12
to reaction-...@googlegroups.com
Attached is a small VTI file containing Tim's 3-D glider, and a slight
modification adding one spot, to travel in a helix. I've only run this
for 1,500,000 timesteps so far, so they haven't collided yet but of
course eventually they will.
I created the gliders in pure rectangles by editing the VTI in Emacs,
then did Generate Initial Pattern and Save, to make a file small
enough to include in a future Ready release.
On 10/4/12, Tim Hutton <tim.h...@gmail.com> wrote:
> This video claims to have one in SmoothLife, which I presume is a
> single chemical:
> http://www.youtube.com/watch?v=7RsEyu8Ol-Q
> But it might have wacky rules, I don't know anything about it.
I corresponded with Stephan Rafler for a while back in March 2012. He
is brilliant and has discovered a lot of stuff in the "smoothlife"
system.
As I recall, the basic idea behind Smoothlife is to generalize
Conway's Game of Life to a continuous domain (both in space and in
time). Each point in space has a value ranging from 100% "alive" to 0%
or "dead". At every moment in time you take a census of the cells in
an "annulus" or "shell" centered on the point (that is, add up the
amount of life-value in all cells whose distance is between K_1 and
K_2). The cell lives or dies depending on its own current state and
upon whether the "census" of neighbors falls within a certain range.
You get different results based on the parameters: the ratio between
the inner and outer radius of the annulus or shell, and the cutoff
values for the census values that determine birth and death. Something
else is done to smooth it out over time, but I've forgotten the
details. It is an extremely compute-intensive calculation (similar to
the many-body gravity problem, e.g. galaxy collisions), and has no
direct analogy to the real world.
Mr. Rafler said he believed that this system was equivalent to
reaction-diffusion systems. It's not immediately obvious to me, but my
guess was that an R-D system could approximate the annulus effect
using multiple chemicals that diffuse at different rates, and a
combination of nonlinear reactions that emulate the birth and death
calculations. (Unfortunately, I offended Mr. Rafler with a
discouraging statement about academic journals, and he fired off a
series of increasingly bitter replies and then broke off
correspondence. So I never got to find out how his discoveries might
relate to R-D.)
The cool thing about Rafler's glider is how closely it resembles the
Schenk et al. "quasi-particles". The Schenk glider looks like an
acorn; Rafler's glider is like the cupule (see
http://en.wikipedia.org/wiki/File:Acorn_diagram.jpg ).
Based on the shape of my 2D "U-skate" glider, I guessed that the
simplest glider in 3-D would look like the cupule of an acorn, or the
bell-shaped part of a jellyfish.
--
Robert Munafo -- mrob.com
Follow me at: gplus.to/mrob - fb.com/mrob27 - twitter.com/mrob_27 -
mrob27.wordpress.com - youtube.com/user/mrob143 - rilybot.blogspot.com
modification adding one spot, to travel in a helix. I've only run this
for 1,500,000 timesteps so far, so they haven't collided yet but of
course eventually they will.
I created the gliders in pure rectangles by editing the VTI in Emacs,
then did Generate Initial Pattern and Save, to make a file small
enough to include in a future Ready release.
On 10/4/12, Tim Hutton <tim.h...@gmail.com> wrote:
> This video claims to have one in SmoothLife, which I presume is a
> single chemical:
> http://www.youtube.com/watch?v=7RsEyu8Ol-Q
> But it might have wacky rules, I don't know anything about it.
I corresponded with Stephan Rafler for a while back in March 2012. He
is brilliant and has discovered a lot of stuff in the "smoothlife"
system.
As I recall, the basic idea behind Smoothlife is to generalize
Conway's Game of Life to a continuous domain (both in space and in
time). Each point in space has a value ranging from 100% "alive" to 0%
or "dead". At every moment in time you take a census of the cells in
an "annulus" or "shell" centered on the point (that is, add up the
amount of life-value in all cells whose distance is between K_1 and
K_2). The cell lives or dies depending on its own current state and
upon whether the "census" of neighbors falls within a certain range.
You get different results based on the parameters: the ratio between
the inner and outer radius of the annulus or shell, and the cutoff
values for the census values that determine birth and death. Something
else is done to smooth it out over time, but I've forgotten the
details. It is an extremely compute-intensive calculation (similar to
the many-body gravity problem, e.g. galaxy collisions), and has no
direct analogy to the real world.
Mr. Rafler said he believed that this system was equivalent to
reaction-diffusion systems. It's not immediately obvious to me, but my
guess was that an R-D system could approximate the annulus effect
using multiple chemicals that diffuse at different rates, and a
combination of nonlinear reactions that emulate the birth and death
calculations. (Unfortunately, I offended Mr. Rafler with a
discouraging statement about academic journals, and he fired off a
series of increasingly bitter replies and then broke off
correspondence. So I never got to find out how his discoveries might
relate to R-D.)
The cool thing about Rafler's glider is how closely it resembles the
Schenk et al. "quasi-particles". The Schenk glider looks like an
acorn; Rafler's glider is like the cupule (see
http://en.wikipedia.org/wiki/File:Acorn_diagram.jpg ).
Based on the shape of my 2D "U-skate" glider, I guessed that the
simplest glider in 3-D would look like the cupule of an acorn, or the
bell-shaped part of a jellyfish.
--
Robert Munafo -- mrob.com
Follow me at: gplus.to/mrob - fb.com/mrob27 - twitter.com/mrob_27 -
mrob27.wordpress.com - youtube.com/user/mrob143 - rilybot.blogspot.com
Tim Hutton
Oct 9, 2012, 7:47:08 AM10/9/12
to reaction-...@googlegroups.com
On 4 October 2012 14:58, Robert Munafo <mro...@gmail.com> wrote:
> Attached is a small VTI file containing Tim's 3-D glider, and a slight
> modification adding one spot, to travel in a helix. I've only run this
> for 1,500,000 timesteps so far, so they haven't collided yet but of
> course eventually they will.
Thanks Robert, I will add this pattern at some point. I love the helix
> Attached is a small VTI file containing Tim's 3-D glider, and a slight
> modification adding one spot, to travel in a helix. I've only run this
> for 1,500,000 timesteps so far, so they haven't collided yet but of
> course eventually they will.
glider. Does it move in a helix though? I can't work it out because
the toroidal wrap-around keeps confusing me.
patterns now, smoothglider and SmoothLifeL. Both are pretty amazing.
They work in Ready 0.4, so if anyone wants to try them out before the
next release you can download them from here:
http://code.google.com/p/reaction-diffusion/source/browse/trunk/Ready#Ready%2FPatterns
His sampling uses anti-aliasing which is the same idea as using a
larger finite-difference stencil, I suppose. His transition functions
use the logistic function for a smooth step. Smoothglider has discrete
time stepping. SmoothLifeL has a variable timestep and gives the same
behavior even for small timesteps, as you'd hope.
His code is faster than ours because he uses FFT for the kernel
convolution, which is a big win for large neighborhoods. Clever stuff!
We're not too far behind though.
Working on the 3D one now.
>
> Mr. Rafler said he believed that this system was equivalent to
> reaction-diffusion systems. It's not immediately obvious to me, but my
> guess was that an R-D system could approximate the annulus effect
> using multiple chemicals that diffuse at different rates, and a
> combination of nonlinear reactions that emulate the birth and death
> calculations.
and the other 'deadness' and they both diffuse then maybe we can work
out the annulus and inner disk fullness from their relative
concentrations at a point. With a single chemical it wouldn't work of
course because there would be no way to distinguish a small nearby
blob from a bigger far-away blob. Worth exploring maybe.
>
> The cool thing about Rafler's glider is how closely it resembles the
> Schenk et al. "quasi-particles". The Schenk glider looks like an
> acorn; Rafler's glider is like the cupule (see
> http://en.wikipedia.org/wiki/File:Acorn_diagram.jpg ).
>
> Based on the shape of my 2D "U-skate" glider, I guessed that the
> simplest glider in 3-D would look like the cupule of an acorn, or the
> bell-shaped part of a jellyfish.
There's a 3D spiral wave in Oreganator that is cupule shaped too. I'll
try and make a pattern for it, but it's quite tricky to get started.
It's interesting that with all of these rules it is a bend that pushes
forwards. Something to do with one chemical getting ahead of another
one, perhaps, as we see most clearly in this video when I was putting
chemical a into the red channel and chemical b into the cyan channel:
http://www.youtube.com/watch?v=-fNI4JftHfo
(No way to get that look in Ready yet, sadly.)
>
> --
> Robert Munafo -- mrob.com
> Follow me at: gplus.to/mrob - fb.com/mrob27 - twitter.com/mrob_27 -
> mrob27.wordpress.com - youtube.com/user/mrob143 - rilybot.blogspot.com
Tim Hutton - http://www.sq3.org.uk - http://profiles.google.com/tim.hutton/
Robert Munafo
Oct 11, 2012, 1:04:17 AM10/11/12
to reaction-...@googlegroups.com
Hi Tim, Sorry for the delay...
On 10/9/12, Tim Hutton <tim.h...@gmail.com> wrote:
> On 4 October 2012 14:58, Robert Munafo <mro...@gmail.com> wrote:
>> Attached is a small VTI file containing Tim's 3-D glider, and a slight
>> modification adding one spot, to travel in a helix. [...]
any symmetry. But I'll prepare a new VTI file containing just that one
glider, and with large enough dimensions that it actually works.
I've been having a bit of trouble manipulating these 3D VTI files
because of the lack of a way to do arbitrary rotation of the pattern
data within the grid, and the limited tools (pencil is much too thin,
paintbrush is about 2x too thick). I pretty much have to edit the raw
ASCII text. I'll probably write automated tools to generate raw ASCII
VTI files at some point in the future.
> Stephan's been helping me implement SmoothLife in Ready. We've got two
> patterns now, smoothglider and SmoothLifeL. Both are pretty amazing.
> They work in Ready 0.4, so if anyone wants to try them out before the
> next release you can download them from here:
> http://code.google.com/p/reaction-diffusion/source/browse/trunk/Ready#Ready%2FPatterns
Those patterns are great. That's exactly the stuff he was showing me
last year. I'm glad you got in touch and got those working!
> Yes, an interesting thought. If one chemical expresses 'livingness'
> and the other 'deadness' and they both diffuse then maybe we can work
> out the annulus and inner disk fullness from their relative
> concentrations at a point. With a single chemical it wouldn't work of
> course because there would be no way to distinguish a small nearby
> blob from a bigger far-away blob. Worth exploring maybe.
Yep... something like that. I thought of doing several chemicals that
diffuse at different rates, allowing more fine-grained detection of
how much living material is at each distance. My mine got bent out of
shape trying to figure it out, so I stopped. (-:
On 10/9/12, Tim Hutton <tim.h...@gmail.com> wrote:
> On 4 October 2012 14:58, Robert Munafo <mro...@gmail.com> wrote:
>> Attached is a small VTI file containing Tim's 3-D glider, and a slight
>
> Thanks Robert, I will add this pattern at some point. I love the helix
> glider. Does it move in a helix though? I can't work it out because
> the toroidal wrap-around keeps confusing me.
It pretty much has to move in a helical path because of its lack of
> Thanks Robert, I will add this pattern at some point. I love the helix
> glider. Does it move in a helix though? I can't work it out because
> the toroidal wrap-around keeps confusing me.
any symmetry. But I'll prepare a new VTI file containing just that one
glider, and with large enough dimensions that it actually works.
I've been having a bit of trouble manipulating these 3D VTI files
because of the lack of a way to do arbitrary rotation of the pattern
data within the grid, and the limited tools (pencil is much too thin,
paintbrush is about 2x too thick). I pretty much have to edit the raw
ASCII text. I'll probably write automated tools to generate raw ASCII
VTI files at some point in the future.
> Stephan's been helping me implement SmoothLife in Ready. We've got two
> patterns now, smoothglider and SmoothLifeL. Both are pretty amazing.
> They work in Ready 0.4, so if anyone wants to try them out before the
> next release you can download them from here:
> http://code.google.com/p/reaction-diffusion/source/browse/trunk/Ready#Ready%2FPatterns
last year. I'm glad you got in touch and got those working!
> Yes, an interesting thought. If one chemical expresses 'livingness'
> and the other 'deadness' and they both diffuse then maybe we can work
> out the annulus and inner disk fullness from their relative
> concentrations at a point. With a single chemical it wouldn't work of
> course because there would be no way to distinguish a small nearby
> blob from a bigger far-away blob. Worth exploring maybe.
diffuse at different rates, allowing more fine-grained detection of
how much living material is at each distance. My mine got bent out of
shape trying to figure it out, so I stopped. (-:
Tim Hutton
Oct 12, 2012, 12:15:17 PM10/12/12
to reaction-...@googlegroups.com
On 4 October 2012 14:58, Robert Munafo <mro...@gmail.com> wrote:
> The cool thing about Rafler's glider is how closely it resembles the
> Schenk et al. "quasi-particles". The Schenk glider looks like an
> acorn; Rafler's glider is like the cupule (see
> http://en.wikipedia.org/wiki/File:Acorn_diagram.jpg ).
>
> Based on the shape of my 2D "U-skate" glider, I guessed that the
> simplest glider in 3-D would look like the cupule of an acorn, or the
> bell-shaped part of a jellyfish.
Attached a render of one of Rafler's 3D gliders, looking very much
> Schenk et al. "quasi-particles". The Schenk glider looks like an
> acorn; Rafler's glider is like the cupule (see
> http://en.wikipedia.org/wiki/File:Acorn_diagram.jpg ).
>
> Based on the shape of my 2D "U-skate" glider, I guessed that the
> simplest glider in 3-D would look like the cupule of an acorn, or the
> bell-shaped part of a jellyfish.
like a U-Skater in cross-section. The vti file that produces this is
committed. It's a bit slow though - only 6fps on my machine.
The SmoothLifeL video I posted went viral. 64000 views on YouTube already.
https://plus.google.com/110214848059767137292/posts/WtPBhYJswAe
Andrew Trevorrow
Oct 12, 2012, 6:19:06 PM10/12/12
to reaction-...@googlegroups.com
Tim:
> The SmoothLifeL video I posted went viral. 64000 views on YouTube already.
> https://plus.google.com/110214848059767137292/posts/WtPBhYJswAe
Brilliant video, and great soundtrack! SmoothLifeL is probably the
most mesmerizingly organic CA I've ever had the pleasure to see.
(Should we do a 0.5 release soon?)
Andrew
> The SmoothLifeL video I posted went viral. 64000 views on YouTube already.
> https://plus.google.com/110214848059767137292/posts/WtPBhYJswAe
most mesmerizingly organic CA I've ever had the pleasure to see.
(Should we do a 0.5 release soon?)
Andrew
Tim Hutton
Oct 12, 2012, 6:24:27 PM10/12/12
to reaction-...@googlegroups.com
On 12 October 2012 23:19, Andrew Trevorrow <and...@trevorrow.com> wrote:
> Tim:
A viral-driven release schedule, that's what we've got. :)
> Tim:
>
>> The SmoothLifeL video I posted went viral. 64000 views on YouTube already.
>> https://plus.google.com/110214848059767137292/posts/WtPBhYJswAe
>
>> The SmoothLifeL video I posted went viral. 64000 views on YouTube already.
>> https://plus.google.com/110214848059767137292/posts/WtPBhYJswAe
>
> Brilliant video, and great soundtrack! SmoothLifeL is probably the
> most mesmerizingly organic CA I've ever had the pleasure to see.
>
> (Should we do a 0.5 release soon?)
Yes, maybe we should. Let me think about it.
> most mesmerizingly organic CA I've ever had the pleasure to see.
>
> (Should we do a 0.5 release soon?)
A viral-driven release schedule, that's what we've got. :)
Reply all
Reply to author
Forward
0 new messages