Permalink Submitted by Maciej Komosinski on Wed, 2003-09-03 14:39
Do you have any pointers to papers where some measures
of symmetry are described for non-directed topological
graphs? (with vertices in a dimensional space)?
Permalink Submitted by Eric Muirhead on Thu, 2003-09-04 08:44
Haven't looked that deeply into it (and your question is WAY over my head
LOL). But I was curious because of the symmetry seen in nature and was
wandering if it can be simulated here as well.
My guesses to nature favoring symmetry over asymmetry would have something
to do with the fact that symmetrical paterns are also the most efficient and
the fractal makeup of life. But I haven't really explored it enough to reply
inteligently on the subject.
"Maciej Komosinski" wrote in message
news:bj5ieb$u4s$1@cancer.cs.put.poznan.pl...
>
> Do you have any pointers to papers where some measures
> of symmetry are described for non-directed topological
> graphs? (with vertices in a dimensional space)?
>
> Or you have your own ideas?
>
>
>
> MacKo
>
Permalink Submitted by Anonymous on Tue, 2003-10-07 09:49
Eric Muirhead wrote:
> Haven't looked that deeply into it (and your question is WAY over my head
> LOL). But I was curious because of the symmetry seen in nature and was
> wandering if it can be simulated here as well.
>
> My guesses to nature favoring symmetry over asymmetry would have something
> to do with the fact that symmetrical paterns are also the most efficient and
> the fractal makeup of life. But I haven't really explored it enough to reply
> inteligently on the subject.
>
>
>
> "Maciej Komosinski" wrote in message
> news:bj5ieb$u4s$1@cancer.cs.put.poznan.pl...
>
>>Do you have any pointers to papers where some measures
>>of symmetry are described for non-directed topological
>>graphs? (with vertices in a dimensional space)?
>>
>>Or you have your own ideas?
>>
>>
>>
>>MacKo
I'm butting in if you don't mind, but Maciej's first suggestion, i.e. a
new genetic encoding would do this most elegantly. If one can implement
something along the lines of these Karl Sims papers, that would go a
long way towards favouring some forms of symmetry:
F4 is something like it, I think, but in my experience doesn't easily
produce limbs spontaneously (maybe this doesn't either, hard to tell
without implementing it.) This may also have to do with the kind of
genetic operations allowed.
Karl Sims' movies of battling creatures were what got me into this
subject, and into framsticks, which I was exstatic to discover about 2
yrs ago, after looking for something like it for years. I would still
like to recreate a framsticks version of that experiment, purely for
fun. As I am not a scientist, and this is pure hobby, it's hard to
find/make the time for this, (I haven't even started on grokking the
scripting-model/language), and I can't say for sure how
easy/hard/impossible implementing this within framsticks would be.
disclaimer: I am not a scientist, and I realize that 1994 is a long time
ago, so it is probable mr. Sims' insights are at least partly surpassed
by the science behind Framsticks and its genetic encodings, I just think
his movies were really cool ;)
Permalink Submitted by Maciej Komosinski on Tue, 2003-10-07 11:39
> I'm butting in if you don't mind, but Maciej's first suggestion, i.e.
> a new genetic encoding would do this most elegantly. (...)
> ...and I
> can't say for sure how easy/hard/impossible implementing this within
> framsticks would be.
It is quite easy, SDK allows for that.
However, I simply don't find it interesting to implement
a genetic encoding which enforces symmetry. What then?
We would get all creatures symmetrical.
What I want to do is to _measure_ how evolution rewards
symmetry, without human influence, and especially, without
genetic enforcement.
Well, it might also be interesting to compare the "standard"
and "symmetrical" encodings... well....... :-)
[I agree that "symmetric" encoding might be useful for
people to design creatures "by hand", if the encoding were SIMPLE.
On the other hand, there is FRED which should simplify
the design of creatures]
> Karl Sims' movies of battling creatures were what got me into this
> subject, and into framsticks, which I was exstatic to discover about
> 2 yrs ago, after looking for something like it for years. I would
> still like to recreate a framsticks version of that experiment...
Yes, such a "coevolutionary" experiment would be nice,
and not very difficult to make in FramScript. Maybe somebody
will do it, or we will in some free moment of time...
Re: A way to increase selection for symmetry? how to measure sym
Do you have any pointers to papers where some measures
of symmetry are described for non-directed topological
graphs? (with vertices in a dimensional space)?
Or you have your own ideas?
MacKo
Re: A way to increase selection for symmetry? how to measure sym
Haven't looked that deeply into it (and your question is WAY over my head
LOL). But I was curious because of the symmetry seen in nature and was
wandering if it can be simulated here as well.
My guesses to nature favoring symmetry over asymmetry would have something
to do with the fact that symmetrical paterns are also the most efficient and
the fractal makeup of life. But I haven't really explored it enough to reply
inteligently on the subject.
"Maciej Komosinski" wrote in message
news:bj5ieb$u4s$1@cancer.cs.put.poznan.pl...
>
> Do you have any pointers to papers where some measures
> of symmetry are described for non-directed topological
> graphs? (with vertices in a dimensional space)?
>
> Or you have your own ideas?
>
>
>
> MacKo
>
Re: A way to increase selection for symmetry? how to measure sym
Eric Muirhead wrote:
> Haven't looked that deeply into it (and your question is WAY over my head
> LOL). But I was curious because of the symmetry seen in nature and was
> wandering if it can be simulated here as well.
>
> My guesses to nature favoring symmetry over asymmetry would have something
> to do with the fact that symmetrical paterns are also the most efficient and
> the fractal makeup of life. But I haven't really explored it enough to reply
> inteligently on the subject.
>
>
>
> "Maciej Komosinski" wrote in message
> news:bj5ieb$u4s$1@cancer.cs.put.poznan.pl...
>
>>Do you have any pointers to papers where some measures
>>of symmetry are described for non-directed topological
>>graphs? (with vertices in a dimensional space)?
>>
>>Or you have your own ideas?
>>
>>
>>
>>MacKo
I'm butting in if you don't mind, but Maciej's first suggestion, i.e. a
new genetic encoding would do this most elegantly. If one can implement
something along the lines of these Karl Sims papers, that would go a
long way towards favouring some forms of symmetry:
http://www.genarts.com/karl/papers/alife94.pdf
http://www.genarts.com/karl/papers/siggraph94.pdf
(Look under the morphology heading.)
F4 is something like it, I think, but in my experience doesn't easily
produce limbs spontaneously (maybe this doesn't either, hard to tell
without implementing it.) This may also have to do with the kind of
genetic operations allowed.
Karl Sims' movies of battling creatures were what got me into this
subject, and into framsticks, which I was exstatic to discover about 2
yrs ago, after looking for something like it for years. I would still
like to recreate a framsticks version of that experiment, purely for
fun. As I am not a scientist, and this is pure hobby, it's hard to
find/make the time for this, (I haven't even started on grokking the
scripting-model/language), and I can't say for sure how
easy/hard/impossible implementing this within framsticks would be.
disclaimer: I am not a scientist, and I realize that 1994 is a long time
ago, so it is probable mr. Sims' insights are at least partly surpassed
by the science behind Framsticks and its genetic encodings, I just think
his movies were really cool ;)
cheers,
eric
> I'm butting in if you don't
> I'm butting in if you don't mind, but Maciej's first suggestion, i.e.
> a new genetic encoding would do this most elegantly. (...)
> ...and I
> can't say for sure how easy/hard/impossible implementing this within
> framsticks would be.
It is quite easy, SDK allows for that.
However, I simply don't find it interesting to implement
a genetic encoding which enforces symmetry. What then?
We would get all creatures symmetrical.
What I want to do is to _measure_ how evolution rewards
symmetry, without human influence, and especially, without
genetic enforcement.
Well, it might also be interesting to compare the "standard"
and "symmetrical" encodings... well....... :-)
[I agree that "symmetric" encoding might be useful for
people to design creatures "by hand", if the encoding were SIMPLE.
On the other hand, there is FRED which should simplify
the design of creatures]
> Karl Sims' movies of battling creatures were what got me into this
> subject, and into framsticks, which I was exstatic to discover about
> 2 yrs ago, after looking for something like it for years. I would
> still like to recreate a framsticks version of that experiment...
Yes, such a "coevolutionary" experiment would be nice,
and not very difficult to make in FramScript. Maybe somebody
will do it, or we will in some free moment of time...
> Just currious if there is a
> Just currious if there is a way to set the parameters so that symmetry is favored.
Either implement another genetic encoding which enforces symmetry, or you can use the measure of symmetry implemented in Framsticks:
http://www.framsticks.com/files/common/NumericalMeasureSymmetry3DCreatur...