Advertisement

Problem in C with NegaMax / Alpha-Beta (Connect 4)

Started by May 23, 2012 06:49 PM
58 comments, last by Jaap85 12 years, 5 months ago
Well, ASM is not portable, so now it becomes important what platform and compiler you are using. This is probably the main reason why you shouldn't use ASM.

Why on Earth did you write parts of this program in assembly language?
biggrin.png i totally agree with you tongue.png
But it's for a project and some parts need to be in Assembly.

I use __asm { } in my C code and it's compiled with VC 2008 so it should be OK for you.



Anyway i can replace ASM parts by C code if you prefer.

It's portable on Windows.
Advertisement
I use __asm { } in my C code and it's compiled with VC 2008 so it should be OK for you.

Except I am not familiar with Visual C, which means more work for me.

Anyway i can replace ASM parts by C code if you prefer.[/quote]
Yes, that shifts the work to you, which is the right place. It will also make your program portable, which is a good thing.

It's portable on Windows.[/quote]
That's a really strange notion of portability.

That's a really strange notion of portability.


Yep but it's because of the cursor positionning and color system (more simple to test it with windows.h on my windows OS and then make it portable).


I gave you the download link (private msg).
Found it. When you are checking for 4 in a line at the beginning of IA_NegaMaxAlphaBeta, you are checking for `joueur', but you should be checking for `joueur^1', because you are checking if the opponent just won in the previous move.
I first stripped the code of anything Windows related and any input or output. I then added a printf statement to show the move and score when a new best move is found at the root. I changed `main' to set up a position and then call the IA once. Once I did that, it took about 2 minutes to produce a position where it wouldn't do the right thing (it wouldn't find a victory after 32 31 30), then 1 minute to run the debugger and find that the victory detection wasn't working.

You should really learn how to use a debugger effectively.

I have many thoughts on the coding style, but I'll save them, since perhaps it's better for you to try to get things working in your own style for a while.
Advertisement
You're right and finally i have to admit i didn't understand at 100% alphabeta, because i haven't been able to debug it myself.
I've had hard times with that, i'm almost ashamed.

But you were a great helper, thanks a lot alvaro, i really appreciate your help and your patience.
Keep going this way smile.png


And by the way... Hope you're a Yankee fan like i am biggrin.png Come on Yankees !
So you have an Italian-sounding name, write your code in French and yet you are a Yankees fan? That's a strange combo. smile.png

I never got into baseball and my wife is a Mets fan.

By the way, the other Belgian with an Italian name I know is Gian Carlo Pascutto, and he probably knows a lot more about alphabeta than I do!

So you have an Italian-sounding name, write your code in French and yet you are a Yankees fan? That's a strange combo. smile.png


Yep, it is a strange mix. But a good one laugh.png


I never got into baseball and my wife is a Mets fan.


Damn... a Mets fan ! Condolences :D


By the way, the other Belgian with an Italian name I know is Gian Carlo Pascutto, and he probably knows a lot more about alphabeta than I do!


I think so. But you are comfortable with alphabeta too, so don't worry smile.png


Thanks again.
Hello,

This might be a bit late, but i finally got my version af a Connect Four AI working (in C#). The link to my code is here:

https://sourceforge.net/projects/connectfour85/

Maybe this can help you out a little bit. My code is currently very simple, without any heuristics or alpha-beta pruning, but it seems to work correctly!

My personal blog on game development!

Black Wolf Game Development

This topic is closed to new replies.

Advertisement