Search the web
Sign In
New User? Sign Up
abalone_prog · Abalone
? Already a member? Sign in to Yahoo!

Yahoo! Groups Tips

Did you know...
Real people. Real stories. See how Yahoo! Groups impacts members worldwide.

Best of Y! Groups

   Check them out and nominate your group.
Having problems with message search? Fill out this form to ensure your group is one of the first to be migrated to the new message search system.

Messages

  Messages Help
Advanced
Re: Abalone Program Test   Message List  
Reply | Forward Message #17 of 96 |
Thanx for trying my lovely abalone ;)

Good to see that i'm fast, it's true that i'm probably not that bad in terms of
search (and plenty of ideas), and still a lot of work to do on evaluation, i
know it ;) Anyway in the version you got, i'm also pruning a lot too much + the
time for pruning decision at each node is too high. i.e. without the futility
stuff i'm at about 500knodes/sec but i evaluate a lot more nodes therefore it
takes more time :) I'm working on it ;)

I'm using the evaluation described in the tele1-02_aich-abalone.pdf paper, with
manhattan/hexagonal distance, but i dont know which weight to give to
barycenter of my stones, my opponent's barycenter, the board center or the
missing stones, and i dont know how to measure the strength (i havent yet
implemented a match betwen my program and itself with different settings and i
cant get the abnet thing work either).

I still have a lot to do it seems :)

David.

Quoting Nacre <nacre_computer@...>:

>
> >[...]
> >I'm interested in getting your opinion about the strength/speed of the
> program
> >(it's my current version, not optimized for everything, lots of clock calls,
> >unoptimized new heuristic ideas i'm still working on). I'm also interested
> in
> >exchanging ideas if you write abalone programs too.
> >
> >
> Hi David.
>
> I tried a game between Nacre 2003 and My Lovely Abalone 1.4. Nacre
> clearly won.
> My Lovely Abalone - Nacre : 0-1
> To me it looked like you have spent a lot of effort on search, and much
> less on evaluation function.You probably have the most advanced Abalone
> program with respect to search, but it looked as you had a simple
> center-distance evaluation function.
>
> If I'm correct, you may be able to leap into top-2 of Abalone programs
> simply by using compactness in evaluation. See also
> http://www.igi.tugraz.at/telematik/tele1-02_aich-abalone.pdf (Werner
> calls it "Geometric Score") or
> http://groups.yahoo.com/group/abalone_prog/message/2
>
> With respect to speed: It doesn't matter (yet). Nacre does 40k nodes/s
> on my computer. My Lovely Abalone 1.4 does 250k nodes/s
>
> Regards,
> Peer
>





Sat Nov 12, 2005 1:15 pm

mogwaifrance
Offline Offline
Send Email Send Email

Forward
Message #17 of 96 |
Expand Messages Author Sort by Date

Thanx for trying my lovely abalone ;) Good to see that i'm fast, it's true that i'm probably not that bad in terms of search (and plenty of ideas), and still a...
david.malek@...
mogwaifrance
Offline Send Email
Nov 12, 2005
1:15 pm

Ok, so I've also tested version 1.5 of My Lovely Abalone against Nacre 2003. This time the results are more mixed. 1 sec/move: MLA-Nacre = 4-6 2 sec/move:...
peer_sommerlund
Offline Send Email
Nov 17, 2005
8:33 pm

Hi Peer, Thank you for the good news, it give me courage in the middle of coding ! I'll try again with abnet, i really wish it to work, but i'm very very very...
david.malek@...
mogwaifrance
Offline Send Email
Nov 18, 2005
8:30 pm

My program now beats abapro 4.8 at champion level for belgian daisy ;) For the time being i'm not slow anymore, Download 1.8 it at...
david.malek@...
mogwaifrance
Offline Send Email
Nov 21, 2005
2:11 pm
Advanced

Copyright © 2009 Yahoo! Inc. All rights reserved.
Privacy Policy - Terms of Service - Guidelines - Help