[Icfp04-discuss] K.I.S.S. Team Ants

[Matt Kearse]

clive at xtra.co.nz wrote:
> I think perhaps you are being a bit modest? :-)

> Your ants look to be "right up there" as far as I can see. I ran them
> against (a) the kiss ants - you win, (b) red team - you win, and (c)
> frictionless bananas - you win. Of course, my simulator might still be
> buggy! Your method of storing the food in the centre of your home looks like
> a winner. We try to surround the enemy anthill and this seems to work pretty
> well against a lot of the other ants but doesn't seem to be very effective
> against yours.

I ran mine against a few other ants from the .zip file someone created
last night and it seemed to win against many of them but there were 3
out of the 10 I tested against that beat me - I can't remember off
hand which ants they were, but Red team was one of them, although on
the occasional map I narrowly beat them.

I have no idea why the strategy of surrounding my anthill doesn't
work. When I read about that idea I thought my ants would easily loose
to it.

I haven't actually written anything about my strategy yet - haven't
had any time since the contest but I will post something when I have.

Here is my ant source file:

http://netwin.co.nz/users/matt@netwin.co.nz/icfp2004/mattk.ant.txt

It is badly documented but it might give you a little insight into
what it does.

Ant the actual ant code generated from that:
http://netwin.co.nz/users/matt@netwin.co.nz/icfp2004/mattk-1.ant
http://netwin.co.nz/users/matt@netwin.co.nz/icfp2004/mattk-2.ant

The only difference between my 2 submissions are a few of the
parameters to flip commands.

Matt.

> G'day Matt

> I think perhaps you are being a bit modest? :-)

> Your ants look to be "right up there" as far as I can see. I ran them
> against (a) the kiss ants - you win, (b) red team - you win, and (c)
> frictionless bananas - you win. Of course, my simulator might still be
> buggy! Your method of storing the food in the centre of your home looks like
> a winner. We try to surround the enemy anthill and this seems to work pretty
> well against a lot of the other ants but doesn't seem to be very effective
> against yours.

> Do you have a write up I can read about your approach, etc.?

> I had a lot of fun but unfortunately had a late night before it started, and
> a late night Saturday, Sunday and Monday also! End result was a lot of very
> low quality "thinking"! Our ants are pretty fast and furious but didn't have
> any really useful way of marking the way home, or to food, etc. I.e. They're
> fast, furious AND random!

> So, go MattK, go NZ! :-)

> Cheers
> Clive
> ----- Original Message -----
> From: "Matt Kearse" <matt at netwin.co.nz>
> To: <icfp04-discuss at lists.seas.upenn.edu>
> Sent: Wednesday, June 09, 2004 8:02 AM
> Subject: Re: [Icfp04-discuss] K.I.S.S. Team Ants


>>> Are there any other New Zealand entries?

>> The DylAntz team was 4/7 NZ.

> I'm also from NZ - I didn't manage to get a team together so entered
> on my own.

> My solutions are at.
> http://netwin.co.nz/users/matt@netwin.co.nz/icfp2004/mattk-1.ant
> http://netwin.co.nz/users/matt@netwin.co.nz/icfp2004/mattk-2.ant

> I ran out of time to add many interesting strategies to them but at
> least they manage to collect all the food on the few sample world maps
> with 1 hive removed that I had time to test with.

> One strategy I did implement which I haven't seen anyone else mention
> yet (I haven't read all the participant pages yet so someone may have
> mentioned it) is to avoid falling into ant traps by checking both
> leftahead and rightahead for a foe ant before making any exploring
> moves. If I see an enemy ant at either location, I move somewhere
> else. This avoids all the static ant traps, but can still fall into
> the clever non-static ant traps that some people came up with.

> With a bit more time I'm sure I could have made my ants perform better
> than they do now.

> Matt.



> _______________________________________________
> Icfp04-discuss mailing list
> Icfp04-discuss at lists.seas.upenn.edu
> http://lists.seas.upenn.edu/mailman/listinfo/icfp04-discuss