Jump to content
  • Advertisement
Sign in to follow this  
DeepButi

Lower bound in zero-window searches

This topic is 5409 days old which is more than the 365 day threshold we allow for new replies. Please post a new topic.

If you intended to correct an error in the post then please contact us.

Recommended Posts

I am testing a cut-off for a constant, cumulative score game. The idea is "even if I would win all remaining points, I have already better things to do".
for (each move)
{
    DoMove();
    CalculateScore();
    if (current_score + remaining_score <= alpha)
    {
        UndoMove();
        ret_value=current_score;
        break;     // cut-off
    }
    NegaScout(....) ... or ... NegaMax(...)
    ....
}
While in most situations this works fine (and it produces an improvement of 20%), sometimes it fails to find the correct solution. I tested it with NegaMax and it works, but NegaScout or MTD(f) (calling NegaMax) fails, so it's clear than the problem is with zero-window calls. The reverse idea "even if I would not make any more points, the opponent will not let me get here" always works
    ...
    if (current_score >= beta)
    {
        ret_value=current_score;
        break;
    }
    ...
As a 20% improvement is really important, I guess if there is a solution, but I cannot figure out how to do it. Any ideas? Thanks

Share this post


Link to post
Share on other sites
Advertisement
Sign in to follow this  

  • Advertisement
×

Important Information

By using GameDev.net, you agree to our community Guidelines, Terms of Use, and Privacy Policy.

GameDev.net is your game development community. Create an account for your GameDev Portfolio and participate in the largest developer community in the games industry.

Sign me up!