Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Correct algorithm for testing move order %

Author: Dann Corbit

Date: 11:21:23 03/05/04

Go up one level in this thread


On March 05, 2004 at 13:06:11, José Carlos wrote:

>On March 05, 2004 at 12:47:50, Andrew Wagner wrote:
>
>>Just checking...to test move order %, which some have said should be around 90%,
>>i'm using this basic algorithm in my main search:
>>
>>GenMoves()
>>for (i=0; i<numMoves; numMoves++)
>>{
>>    MakeMove(Move[i]);
>>    val = -search(depth-1,-beta,-alpha)
>>    if (val>=beta)
>>    {
>>        return beta
>>    }
>>    if (val >alpha)
>>    {
>>       if i = 0
>>       {
>>            FirstMoveFailHigh++
>>       }
>>       FailHigh++
>>         .
>>         .
>>         .
>>    }
>>}
>>return alpha
>>
>>
>>
>>Then the number i'm using for move order % is (FirstMoveFailHigh/FailHigh)*100
>>
>>Can someone verify for me that i'm testing this correctly?
>
>  Fail high means val >= beta, so you're currently counting exact values, not
>fail highs.

Maybe he uses alpha and beta in the opposite sense like Ed does.



This page took 0 seconds to execute

Last modified: Thu, 15 Apr 21 08:11:13 -0700

Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.