Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Crafty 17.8 question

Author: Robert Hyatt

Date: 05:44:35 03/01/00

Go up one level in this thread


On March 01, 2000 at 04:52:42, Tim wrote:

>Hello, Dr. Hyatt.
>I am talking about when presult is 2 not 1.
>
>Crafty returns 2 when next condition meets during pondering.
>
>       "pondering was done, but the ponder search
>       terminated due to either finding a mate, or the
>       maximum search depth was reached.  the result of
>       this ponder search are valid, but only if the
>       opponent makes the correct (predicted) move."
>
>
>But you didn't assign value = last_root_value when pondering result was 2.
>
>    if (presult == 2) {
>      if((From(ponder_move) == From(move)) &&
>         (To(ponder_move) == To(move)) &&
>         (Piece(ponder_move) == Piece(move)) &&
>         (Captured(ponder_move) == Captured(move)) &&
>         (Promote(ponder_move) == Promote(move))) {
>        value=last_root_value; <==== Didn't you forget to add this line?
>        presult=1;
>        if (!book_move) predicted++;
>      }
>      else presult=0;
>    }
>    ponder_move=0;
>    thinking=1;
>    if (presult != 1) {
>      strcpy(whisper_text,"n/a");
>      last_pv.pathd=0;
>      last_pv.pathl=0;
>      display=tree->pos;
>      value=Iterate(wtm,think,0);
>    }
>    last_pv=tree->pv[0];
>    last_value=value;
>    if (abs(last_value) > (MATE-300)) last_mate_score=last_value;
>    thinking=0;
>    if (!last_pv.pathl) {
>      if (value == -MATE+1) {  <======= This line is not executed when
>					pondering result was 2.
>					Because you didn't assign value.
>        over=1;
>        if(wtm) {
>          Print(4095,"0-1 {Black mates}\n");
>          strcpy(pgn_result,"0-1");
>        }
>        else {
>          Print(4095,"1-0 {White mates}\n");
>          strcpy(pgn_result,"1-0");
>        }
>      }
>
>I added value=last_root_value. But I don't think it works in every cases.
>
>Thank you.


this is fixed elsewhere to make the code easier to understand.  the next version
of the source will be out soon and fixes the 'position learning' problem which
was not related to the above (the function was called incorrectly).



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.