Framsticks

Hello, all users!

I name me as a newbie, as I have framsticks since a month or so. I was able
to create creatures very soon, but they didn't evolute into something
worthfull. Here an example:

My starting creature was:

MMMWWWWXMMMWWWWX(,,WWWWX[S],WWWWX[S],,,WWWWXWWWWX,MMMWWWWX[N,-2:1,-1:1,0:1,1
:1][|-1:-.8][|0:.1,4:.8]MMMX[|-1:.8]MMMX[|-1:.8]MMMX[|-1:.8]MMMX[|-1:.8]MMMX
[|-1:.8]X[|-1:.8]RRMMMX(llllX,XX))

After about 6 Mio. steps I got something like that:

X(X[S],MX[N,0:2.544][N,0:.8](llllX(,,X[|-1:.5,0:1.136]WX[|-1:6.925,0:-1.133]
WX[|-1:0.888,0:.5]RRX[|-1:.5,0:-1.793](llllX,X)),WX[|-7:.1]X[|-1:.1]XRRXWX[|
-1:2.071,0:.5](X[@,-5:1]lllX[|-2:.1]XRRXI(lX),,)))

I used the following parameters:

Parameters: Gene pool capacity: 200; Delete genotypes: only the worst;
Simulated Creatures:3; Creation height:0

Selection: Unchanged:20; Mutated: 60; Crossed over: 2; Minimal similarity:
0; Selection rule: Tournament (2 genotypes)

Fitness: Costant: 0; Life span: 10000; Velocity: 10000; Body parts: 5; Body
joints: 5; Brain neurons: 10000; Brain connections: 10000; Distance: 10000;
Vertical position: 0; Vertical velocity: 0; Criteria normalization: no;
Similarity speciation: no

Energy: Starting energy: 2000; Idle metabolism: 1; Automatic feeding: 30;
foods energy: 200; foods genotype: no; Ingestion multiplyer: 1; Aging time:
0

Gene pool: Genotypes: Fitnes formula: return 0.0+this.distance*10000.0;
Scale fitness: yes; Shift coefficient: 1; Scaling coefficient: 1

The other datas were unchanged from normality.

My world is: Type: Flat surface; Size: 70; Water level: 20; Boundaries: none

Here you have my complete experiment. Can anyone help me? What did I do
wrong? As you see, my intention is to create "animals" swimming around and
searching for food. But every creature only drifts through the water, head
up and tail down. That's it, nothing moves.

Ernie from Germany

hi... i tried the speed.sim parameters, but the mutation rate was too
little... i increased them a lot, calibrated them, and starting with a
bacteria, i got a creature with 0.142251 as velocity! if u want its code, it
is:

Siw(rLLLAAAffsIwEF(,,(LLLFSIX[LMu,2:0.289]A(LLLafMMMMMMMIIIIIeeeX[*][@1:1.09
4, si:2][@=:1.617, 0:4.47,si:2.54218],aaafmSQ(LLFFFFMFFFMMMMMMqqqX[D,
1:2.297, -2:-1.344][|-2:0.861, 2:-1.312, s:0](rfEsqEq(aCCfmmWWWX[N,
2:5.141,-3:1][@-2:-1.352, in:0.8, s:-0.307, -3:1][|!:0.27076, -1:0.912,
si:999, in:0.81408]))),), , )), )

totaly envolved with my crazy, high-mutating parameters, from a simple
bacteria (X).

Ahh, and i have a question to the developers:

when i came back home, there was a "Floating Point Error" window. When i
clicked ok, framsticks closed, but, thank god, i didn't lose my gene pool.
but i lost propably a LOT of time...

[]'s

bruno

hi... i'm trying to envolve fast creatures... what are the best parameters
for this? (please, tell me the whole configuration, since i'm kinda newbie)

ah, and is 22.5kSteps/sec a good speed?

another thing: how to envolve fast food searchers?

thanks

bruno

Is there any way to implement different methds of selecting for fitness
without getting into C++?

I am interested in trying to implement the FUSS method of selecting
individuals according to fitness, if it is possible. It is a method that
attempts to preserve genetic diversity.
ftp://ftp.idsia.ch/pub/techrep/IDSIA-01-0.ps.gz &
http://www.idsia.ch/~marcus

Hi I am trying to get the 2.2/3 versions to get 10 creatures and 20 food
individuals, however I can only get the creatures population to calculate,
also
I cannot select the number of individuals for food either (I can select the
number for the creatures population as in version 1)

I thought the idea was to have both populations compete, any clues?

--
Ian Mackenzie

2F1 2 Lord Russell Place, Edinburgh EH9 1NQ

Tel 0131 667 9916 (BT), 0131 477 6324 (Cable), 07780814717 (Mobile)

my windows wont let me install framstiks because it requires version
4.0.1111 I looked at my version and it is 4.00.950a
how can I make it work? I cant find an update.

hello

I do not manage to have creatures which are nourished.
I have many energie balls in the world but they never eat them.
I have activated the parameter "life span" but that not changes anything.
I cant find the parameter "Eatable corpse" . Is there in version 2.3 ?
Thank you for your assistance

Could it be possible to introduce sensors that look for food only, or to
other 'living' creatures only?

I extended my food finding experiments into a competition, with each time
having two individuals compete for three food balls.
The smell sensors do not distinguish between other creatures and food, so
many times the individuals just keep facing and repelling each-other,
without going for the food laying around, and eventually both starving.
(Frams ARE very much like humans!)

One way to do this would be to build a custom neuron, scanning the
LiveLibraries and providing an answer. I am going down that route for now,
but it is considered that this polling will be a rather slow process to
build in a neuron. 'Hard coded' would be much faster.

Just my suggestion for version 2.4.

Frans Verbaas
frans.verbaas@freeler.nl

I'm making an experiment, but all the time a message beeps, saying (in that
messages window, that opens automaticaly on the message if i close it)
:onBorn - couldn't avoid collisions!

i already removed the beep, setting the beep to at least critical errors...
but that windows slows down the evolution...

can any1 help me?

another thing, on an Athlon 1.2ghz 256mb ram on win98, simulating 5
creatures at a time, in the "F9" screen mode, i am having about 3300
steps/sec... is it good?

thanks!

bruno

Hi

Is there a top speed record? What is it? And how was it achieved, I mean
what circumstances, What special settings are needed to make sure that all
the velocity measurements are equal?

You hear a lot about a "performance sampling rate", but what on earth is it,
where do I find it to change it, and what should it be/should I change it to
for my velocity results to be accurate?

Thanks alot
Grant