November cw tourney
Nenad Tomasev
As I've already said before, I plan on organizing a tourney that should
start soon, so I'd like all of those who are interested in
participating to reply to this post, so that I could soon find out the
number of participants, which will be crucial in determining the
competition system. Basically, it should be similar to CSEC, but we'll
see. :)
The starting date is still flexible, but let's say for now that the
competition starts 15.11.2006. , and you should say that you want to
take part before 13.11.2006.
I promise that it will be interesting :) the first round or two may be
a little bit more conventional, but expect some crazy stuff along the
way :).
Cheers.
Nenad
CoreChild
John
inversed
I'll participate anyway, but I'd prefer scoring system like in The
Oversby Autumn 1998 Tournament.
inversed
will
/Will
----- Original Message ----
From: Nenad Tomasev <tom...@nspoint.net>
> The starting date is still flexible, but let's say for now that the
> competition starts 15.11.2006. , and you should say that you
> want to take part before 13.11.2006.
Send instant messages to your online friends http://uk.messenger.yahoo.com
fatalc
though.
Fluffy
you can count me in, too :-) But as others have already remarked
a new scoring system would be nice - a system where
everybody can take part in every round. And don't forget to think
about people who can't participate in each round.
Fluffy
Roy
How about no elimination but a overall score for all round, like the
frenzy did. Or maybe a score system like the Formula 1, every round the
winner gets 10 points, the second place 9...etc.
Ciao
saye...@gmail.com
p/s: On top of this, I have a suspended tournament I guess...
Łukasz Adamowski
> I'm in too!
>
> How about no elimination but a overall score for all round, like the
> frenzy did. Or maybe a score system like the Formula 1, every round the
> winner gets 10 points, the second place 9...etc.
I'm in and I like this scoring system.
Or why not to continue Redcoders' Frenzy? It's seems to be quite dead right now, but has well knonw website (a bit out of date, but it can be fixed). The scoring system used in RF is also nice.
Lukasz Adamowski
-= "War is a problem, never a solution" =-
-= "Wojna jest problemem, a nie rozwiazaniem" =-
----------------------------------------------------
Własne M3? Kup zanim będzie drożej! Raty od 300,- zł Kliknij:
http://klik.wp.pl/?adr=www.kredyty.wp.pl%2Fkredyty.html&sid=906
Cristian210273
> start soon, so I'd like all of those who are interested in
> participating to reply to this post, so that I could soon find out the
> number of participants, which will be crucial in determining the
> competition system. Basically, it should be similar to CSEC, but we'll
> see. :)
with an evolved warrior for the nano-hill!!!!
So the question arises..... which is the permitted length???
To realize a longer warrior (to be competitive for higher
length) he need a little bit more time, so how about the
beginning of December?
Chip
I like Roy's Formula 1 scoring idea. Can we also increase the maximum
number of points each round (maybe the winner of the first round gets
10 points, the winner of the second round 12, then 14,...), so that
even people who are behind have a chance to catch up?
- Chip
Sascha Zapf
I would prefer Roy's scoring system, or better Chips idea
Sascha
------------------
Powered by NetMail
Roy
will help new programmers more...
If you are last for 5 rounds in a row and get some points in the 6th,
with everybody getting points you are far behind and don't climb. But
if more people where stuck at 0 points, you can rise much faster in the
ranking.
Maybe we should even give out less points so it stays unpredictable,
like in F1 we should keep the points +/- 50% of the contestants. Even
in rounds with less people sending warriors (easier round to win/score)
you then get less points.
I propose a scoring system of:
nr 1 scores AMOUNT_OF_CONTESTANTS / 2 (= Z)
nr 2 scores Z - 1
nr 3 scores Z - 2
nr N scores Z - N
(Negative numbers become 0)
Ciao!
CoreChild
> nr 2 scores Z - 1
> nr 3 scores Z - 2
> nr N scores Z - N
> (Negative numbers become 0)
How about negative numbers (and zero) become 1, so at
least credit is given for taking part?
John
will
could you up the points by creating complete facades of players who just happen to submit warriors that succumb most to your own? ;-)
Hmm, maybe that works even when the scores aren't #contestants-based
/Will
----- Original Message ----
From: Roy <roy.va...@gmail.com>
Subject: Re: November cw tourney
I propose a scoring system of:
nr 1 scores AMOUNT_OF_CONTESTANTS / 2 (= Z)
nr 2 scores Z - 1
nr 3 scores Z - 2
nr N scores Z - N
(Negative numbers become 0)
Cristian210273
Where can I found the rules to evolve a warrior for this
tournament?!?!?!
For "rules" I mean, for instance:
- core length
- max no. of cycles
- max no. of instructions
- ....
Sorry if the question sounds silly, but I'm a newbie of CW......
fizmo
later.
BTW, I will participate too :)
Chris
Fluffy
Cristian210273 wrote:
> Where can I found the rules to evolve a warrior for this
> tournament?!?!?!
> For "rules" I mean, for instance:
> - core length
> - max no. of cycles
> - max no. of instructions
> - ....
usually the rules will be published only a week or so before
the start of the tournament.
> Sorry if the question sounds silly, but I'm a newbie of CW......
Just keep on asking your "silly" questions :-) You might also consider
asking your questions on #corewars at irc.koth.org. That way you might
get a response faster.
- Fluffy :-)
Cristian210273
> usually the rules will be published only a week or so before
> the start of the tournament.
>
I hope to have time to evolve my warrior for the tournament,
provided that evolved warriors will be accepted!!!!!
saye...@gmail.com
round. evolver gets lazy when dealing with big cores e.g. 8000 cores or
more. Even for 800-core, evolvers already showing tendency to stuck at
local maxima, usually far away from global maxima
I hope it is the algorithm, not the concept weakness....
Cristian210273
> round. evolver gets lazy when dealing with big cores e.g. 8000 cores or
> more. Even for 800-core, evolvers already showing tendency to stuck at
> local maxima, usually far away from global maxima
Ok, guys, now I'm really confused!!!!!
Does the preceding discussion mean that each player in the tournament
shall send a warrior for each round he has successfully
completed?!?!?!?
Is not a single warrior sufficient for all the tournament's rounds??????
ach...@tin.it
saye...@gmail.com
Welcome to the tourney :)
Robert Macrae
It is the nature of the problem, not of the algorithm. GAs are not
particularly vulnerable to local minima, CW is such a horrible space in
which to optimise (discontinuous, rough, deep and narrow minima surrounded
by flat plains) that any algorithm will suffer.
Robert
saye...@gmail.com
Randy Graham
<cristia...@yahoo.it> wrote:
Correct. Every round has new rules, and therefore every round tends
to have new warriors. In the past, we have had contests which
included sorting data in a preset location in core, submitting 2
warriors which cooperated to defeat the opponent's two warriors,
defeat an opponent without damaging a 3rd neutral non-attack program,
straight up 1 on 1 fights, fights within randomly varying cores, and
so on. The amount of stuff do-able within corewars is actually
surprisingly complex, and many of the tournaments cover broad aspects
of the game over many rounds.
RagManX
--
http://www.blahblahblahg.com/ - Blah about me
http://www.gamepatches.info/ - Just like it says
Łukasz Adamowski
Hopefully. ;]
BTW I've written an evolver some time ago. It's pretty slow because of using a lot of files (hard drives are slow, you know) and I used it only once optimizing constants. And you know what? I still prefer hand-coding. ;]
Greetings
Lukasz Adamowski
-= "War is a problem, never a solution" =-
-= "Wojna jest problemem, a nie rozwiazaniem" =-
----------------------------------------------------
Nowa Gra Zakręcone Dynie - sprwdź i kliknij:
http://klik.wp.pl/?adr=http%3A%2F%2Fadv.reklama.wp.pl%2Fas%2Fd24.html&sid=911
Czy zakręcisz nimi jeszcze bardziej i zbijesz wszystkie z planszy?
Cristian210273
> Hopefully. ;]
Uhmmmm...... As I had already pointed out, I'm a newbie so my
considerations
could be silly! But.... I would modify the preceding sentence in:
"..... why evolvers have NOT YET dominated all the hill......"
>From an EC fans ! :-)
However I think that we have to discriminate between EVOLVED WARRIORS
and
AUTOMATICALLY GENERATED WARRIORS (evolved warriors could be considered
a subset of automatically generated warriors). Anyone knows some
application,
different from Evolutionary Computation designed for Core Wars? I'm
planning
to explore a new tecnique, and I would like to know if someone else has
already
explored the field....
Regards!
saye...@gmail.com
second your opinion :)
Apart from EC, I have not read anything else regarding corewar warrior
'auto' generation. Is there anything else that auto-generating? How
could I miss that elegant concept.
Cristian210273
> Apart from EC, I have not read anything else regarding corewar warrior
> 'auto' generation. Is there anything else that auto-generating? How
> could I miss that elegant concept.
>
to tell my idea: the strategy is UNDISCLOSED!
:-)
However I have a basic question:
provided that I'm able to design a program that auto-generates a
good warrior with no use of EC, in which hill could I submit this
warrior????
Regards!
saye...@gmail.com
Cristian210273 wrote:
> However I have a basic question:
>
> provided that I'm able to design a program that auto-generates a
> good warrior with no use of EC, in which hill could I submit this
> warrior????
>
You can skip nano. Leave it to existing evolvers.
The traditional test ground, or benchmark, for non-human created
warriors is tiny hill. Start your battle here.
If you are still need more challenge, proceed to climb 88 or 94 hills
(8000 cores).
Advance to huge core (55440) once the 8000 cores have been conquered.
Nenad Tomasev
Hello once again :)
It's almost time for the tournament to start, so I thought I should
post some details here.
It won't be an elimination tournament. It will most likely be like a
set of RF rounds put together. As for the scoring system, it'll be like
someone here mentioned - 10 points for the winner of the round, 8 for
the second, and 6, 5, 4, ... 1 for the rest of the top 8 players.
Nothing unusual, but I think that it'll work just fine, anyway :).
Two weeks will be the time given for each round. I think that it'll be
just enough for those with other things to do to be able to find the
time to make their warriors, but not too much either.
There will probably be 8 rounds. (I say "probably", because it also
depends on your interest to play, and I'd like for all of you to say if
you'd like to play more or less) This issue will be resolved by the end
of the first round. The first round, as usual, won't be difficult to
play... a kind of a warm-up.
This is the preliminary list of participants (there is still time for
someone else to join the tournament):
John Metcalf
LAchi
Robert Macrae
inversed
Will Varfar
fatalc
Jens Gutzeit
Roy van Rijn
Zul Nadzri
Lukasz Adamowski
Chip Wendel
Sascha Zapf
Christian Schmidt
Mizcu
Christian 210273
Phil
ia...@admin.drake.edu
> someone else to join the tournament):
I'm in.
P. Kline