[Icfp04-discuss] Can someone beat this.....
Jens Hauke
hauke at ipd.uka.de
Tue Jun 8 13:40:12 EDT 2004
Hi,
We (KaNT) also tryed some GA (used for lightning submission). The ants
find food, but we did'nt start the GA from scratch. The 150-x.ant's are
interesting, because they found themself a solution for "home defence"
with high enemy killing rate after 2 days of evolution! They kill more
than 50 ants of our final solution (but than still lose).
The ants:
http://www.ipd.uka.de/~hauke/icfp2004/
The language for the handcoded ants:
http://www.ipd.uka.de/~hauma/icfp2004/
I think, we also spent to much time for writing the compiler, the
simulator and the simulator gui - not enough time left to improve ant
strategy's.
It's amazing, what is possible in just 72h.
Thanks to the organisers! Great Task!
Jens
Jin Soo Hwang wrote:
> christian sloma schrieb:
>
>> I think i know your feelings ;)
>> my GA was finally running at 6 this morning ( 12 hours left)
>> and i havent found any ant that performs better than the random
>> searching ant given in the task description until now ;((
>> (had to restart severall times, bc of some smaller bugs too, im now in
>> generation 500+, still trying...)
>
> *gg* like me. the ants I have submitted was the 20th generation. But now
> after certain rounds I have ants which collect foods, what a nice
> suprise. Too late but what a joy. :-)
>
More information about the Icfp04-discuss
mailing list