Tic Tac Toe Coding Challenge

[cosmin]

Hi,

Bruce ideea is excellent, so anyone can see the mystakes that were made, maybe i made some mystakes with my tester (when i get some time, i will lookk again in it). Since Bruce will post all entries, i will not post my sol.

cosmin

[BruceAmmons]

Charly and I had a discussion, and we realized his Aikido player had a bug in it.  He sent me a new version of the player, and it beat everything else easily.  With just 100 matches between each player, the Aikido player scored 4407, and the next closest player was Adrien again with 4209.  This is a pretty impressive player.

Since the debugged version was submitted after the contest was over, it won't change the official results.  It is interesting that Charly won anyway, just not with the player he expected.

Bruce

Message Edited by Bruce Ammons on 07-22-2006 01:53 PM

[cosmin]

hi,

god damm a found the error?????????????, it's in my player, i have a vi with is called my recursion, ttc_recursion3.vi, but in the code in the Open VI Reference.vi, the path is wired ttc_recursion.vi, so is not working (in the rush of things is apparent i put the number 3 at the end of the name?????). Bruce , please test my player by open the ttc_recursion3.vi and put the number 3 at the end of input to Open VI Reference.vi and browse again to the same vi in the input type specifier (not to get the connector pane error).

cosmin

[ien_fr]

Hi all, I'm adrien

I send here my bruce submission wich you can use into your tester.

In your tester use RoxallMain.vi

bye

[chilly_charly]

Adrien

Thanks for making your vi public.

Remember to adopt the reference connector pane provided by Bruce, otherwise your player can't be tested automatically.

Now I can further optimize my own player... 😉

[chilly_charly]

Attached is the player (LV 7.1) I used to develop all my different players. I hope the comments will be enough to exposed the adopted strategy.

Open the CC test globals to play with the evaluation grid settings.

I suppose a better optimized grid will be developped once all the players have been disclosed,

Message Edité par chilly charly le 07-23-2006 04:10 PM

[BruceAmmons]

Cosmin,

I tried the fixed version of your program.  It runs much, much slower but it would have won.  It even beat Charly's fixed Aikido player.  I only ran 10 matches against each player, but your player scored 425.  Charly's fixed Aikido player was second with 420.  I'm not sure if this is a large enough sample, but I didn't want to wait any longer.

Sorry that your bug wasn't found before the competition officially ended.  I guess you will just have to be happy with the knowledge that it would have won.

Bruce

[cosmin]

Bruce,

Thanks for taking your time to fix my dumb mistake (I feel very silly about it), just as was said before, is hard to get the time to test everything. I can’t even remember when I put that 3 in the name of vi. I put in the VI some sort of level, so it plays slow because I choose some high level, actually it can play much faster without losing a lot of strength.

Thanks again for your effort, you are a very kind man

cosmin

[cosmin]

Bruce,

From now on, I will read my yahoo mail every day, you dicovered that is something wrong with my player immediately.

cosmin

[altenbach]

Guys,

I noticed that I never showed my code of the TTT trainer that I only provide as an executable long ago.

http://forums.ni.com/ni/board/message?board.id=170&message.id=184759#M184759

For those who are interested, here's one of the latest incarnations from last August. It is still in LabVIEW 7.1. Unfortunatly, I no longer have 7.1 installed, else I would have added a few comments for those interested. (I don't need them because the code is pretty obvious to me ;))

Of course the scoring tables were generated elsewhere and only show as diagram constant.

Enjoy! 🙂

(I am sure it is still buggier than an ant farm. Try to find all the bugs ;))