Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Faster, deeper and more of such...

Author: Dann Corbit

Date: 11:03:50 09/14/00

Go up one level in this thread


On September 14, 2000 at 13:42:49, Ed Schröder wrote:

>In the upcoming Rebel Century 3.0 I have implemented a little
>statistic routine that reveals something about the nature of SEARCH
>that could be important for the future of computer chess in the sense
>that it says "something" one may expect in the near future because
>of faster and faster PC's.
>
>Research on this issue have already been done by Bob and Ernst and
>it has made me curious so I have spend a little time on it. The
>statistic shows 2 things:
>
>a) number of fail-low's for each depth;
>b) number of "changed moves" for each depth.
>
>(a) is not so important as often fail-low's do not mean anything but I
>wanted to know anyway.
>
>(b) is extremely important as it shows for each depth how many times
>Rebel changed its mind. As you can see in the below statistic the %
>diminish and diminish the deeper Rebel goes.
>
>How to read the overview:
>- first column: iteration depth;
>- second column: number of times the depth was reached;
>- third column: number of fail-low's;
>- fourth column: percentage of fail-low's;
>- Fifth column: number of changed moves;
>- Last column: percentage of changed moves;
>
>SEARCH OVERVIEW
>===============
>
>1      4726         0 ( 0)        0 ( 0)
>2      4726      1000 (21)     1889 (39)
>3      4726       495 (10)     1468 (31)
>4      4719       209 ( 4)     1085 (22)
>5      4719       218 ( 4)     1222 (25)
>6      4699       191 ( 4)     1139 (24)
>7      4655       141 ( 3)      948 (20)
>8      4572       109 ( 2)      837 (18)
>9      4457        79 ( 1)      777 (17)
>10      3998        86 ( 2)      644 (16)
>11      3015        64 ( 2)      374 (12)
>12      1904        55 ( 2)      204 (10)
>13      1093        37 ( 3)       77 ( 7)
>14       584        22 ( 3)       35 ( 5)
>15       356        15 ( 4)       22 ( 6)
>16       230         7 ( 3)        6 ( 2)
>17       157         6 ( 3)        2 ( 1)
>18       123         6 ( 4)        3 ( 2)
>19        88         0 ( 0)        0 ( 0)
>20        67         0 ( 0)        1 ( 1)
>21        55         0 ( 0)        1 ( 1)
>22        54         0 ( 0)        1 ( 1)
>23        50         0 ( 0)        0 ( 0)
>24        47         0 ( 0)        0 ( 0)
>25        40         0 ( 0)        0 ( 0)
>26        30         0 ( 0)        0 ( 0)
>27        28         0 ( 0)        0 ( 0)
>28        22         0 ( 0)        0 ( 0)
>29        19         0 ( 0)        0 ( 0)
>30        14         0 ( 0)        0 ( 0)
>31        14         0 ( 0)        0 ( 0)
>32        13         0 ( 0)        0 ( 0)
>33        12         0 ( 0)        0 ( 0)
>34        12         0 ( 0)        0 ( 0)
>35        10         0 ( 0)        0 ( 0)
>36        10         0 ( 0)        0 ( 0)
>37        10         0 ( 0)        0 ( 0)
>38        10         0 ( 0)        0 ( 0)
>39        10         0 ( 0)        0 ( 0)
>40         8         0 ( 0)        0 ( 0)
>41         8         0 ( 0)        0 ( 0)
>42         5         0 ( 0)        0 ( 0)
>43         4         0 ( 0)        0 ( 0)
>44         3         0 ( 0)        0 ( 0)
>45         3         0 ( 0)        0 ( 0)
>46         2         0 ( 0)        0 ( 0)
>47         2         0 ( 0)        0 ( 0)
>48         2         0 ( 0)        0 ( 0)
>49         2         0 ( 0)        0 ( 0)
>50         2         0 ( 0)        0 ( 0)
>51         2         0 ( 0)        0 ( 0)
>52         2         0 ( 0)        0 ( 0)
>53         2         0 ( 0)        0 ( 0)
>54         2         0 ( 0)        0 ( 0)
>55         2         0 ( 0)        0 ( 0)
>56         2         0 ( 0)        0 ( 0)
>57         2         0 ( 0)        0 ( 0)
>58         2         0 ( 0)        0 ( 0)
>59         2         0 ( 0)        0 ( 0)
>60         2         0 ( 0)        0 ( 0)
>
>After iteration 19 one hardly sees any changes. One might wonder
>if a doubling in speed is still good for 50-70 elo (as often said
>here).
>
>One thing that should be added to the overview is a division in
>middle-game and end-game (I think).
>
>The overview was created by playing a 40/40 auto232 match (about
>50 games). Results are automatically kept so that a next auto232
>match the statistic is automatically updated.
>
>Also if you are playing your normal (manual) games or analyze positions
>the statistic is maintained. It would be nice to see how the statistic
>would look like after say 100,000 moves.

I think this is a fantastic idea.  When analyzing an EPD position, I think
"depth of last change" might be nice for a comment (c3 or something).  This
would be especially helpful to know when a position might trip up an opponent.

I disagree that we can form an opinion about the value of deeper searches at
this point.  The reason I make a bold and silly statement like that despite your
clear evidence is the sort of searches that will give 60 plies depth.  If we
were able to achieve 60 plies from the opening position, I think that is far
more valuable than 60 plies in the deep endgame.  I suspect that none of the
positions searched very deeply were complicated board positions with many
possiblilities for alternatives.  When we are in the opening, middle game, or
early endgame, I think these are the places where deep searches are valuable.
If you have only 7 pieces left on the board, there are not that many possibly
good choices.

Of course, I could be completely wrong.  But I think another valuable column
would be "piece count" so that we could do queries based not only on depth but
also upon complication.

Your ideas are very inventive and creative.  No wonder you produce such a
wonderful product.

-- Dann "Slavering for a Win32 EPD analyzer version of Rebel" Corbit



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.