Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Deep Blue-Kasparov 1997 game 2 : positions to be analysed (to Uri Blass)

Author: Robert Hyatt

Date: 10:01:17 09/07/01

Go up one level in this thread


On September 07, 2001 at 09:04:51, Erkki Malkamaki wrote:

>
>Move 22 Deep Blue played b4. Move was first considered at T=60s
>[D]r2qrbk1/3b1ppn/p2p3p/1ppPp3/P3P3/1PP1B1NP/2B1QPP1/R3R1K1 w
>

Crafty likes axb5 until depth=12.  There it changes to Nf5, which
takes about a minute or so to find.  For the first minute then, you would
get axb5.  If it has more than a minute to search, then it will play Nf5.  I
didn't get b4 in any reasonable time.  It did appear that b4 seems to be
reasonable, as it is #2 on the move list (ordered by total nodes searched)
at the start of ply 13, but it didn't replace Nf5.  At the end of ply 13,
crafty switched back to axb5 again.  Score is +.61...  After ply 14 has
finished, taking almost 5 minutes total, axb5 is still best, +.57.

According to my copy of the DB log for this game, it found b4 at time 70,
depth 12 [7(5)].  And it stuck with it thru depth 16.

>
>Move 23  Rec1   T=193s
>[D]r3rbk1/2qb1ppn/p2p3p/1ppPp3/PP2P3/2P1B1NP/2B1QPP1/R3R1K1 w
>


Here early searches like bxc5 with a score of between +.8 and +.9.  This
lasts through ply 12 at least, which takes less than a minute of total search
time to complete.  Depth 13 takes 2 minutes, score is +.72, best move is still
bxc5.  depth=14 score for bxc5 is +.66...  best move stays bxc5 for at least 10
minutes total time.


DB seems to have liked axb5 until it hit Rec1 on a small fail high during
depth 15 [9(6)].  I can't tell when it found this move, but once it did,
it didn't waver thru the next search (16 plies).  It found it prior to 193
seconds, as that is how long it took to complete the whole search.


>
>Move 29  Qf2  T=149s
>[D]r1rqnbk1/3b1pp1/p6p/1p1Pp3/PPp1P3/R1P1B1NP/2B3P1/R4QK1 w
>

Here I get Qf2 instantly, then it changes to Qf3 and then axb5.  Depth 11 takes
less than 30 seconds, score is +.87, best move is axb5.  at depth 12, axb5 is
+.85, and Crafty changes to Qf2, +.87.  At depth=13 Qf2 is still +.87 but it
then switches to R3a2 with a score of +.94.  At 5 minutes it hasn't changed its
mind further.



Here DB took 53 seconds to complete the 15 ply search and produce the move
Qf2.  It stuck with this move thru 17 plies [11(6)].


>Move 33 Nf5  T=182s
>[D]r1r1qbk1/3b1pp1/p2n3p/1pBPp3/PPp1P3/2P3NP/R1B2QP1/R5K1 w


Early searches like Qf3 with a score of +.91... this continues through depth=12.
at depth=13, it changes from Qf3 (+.94) to axb5 (+1.01).

Nf5 happened somewhere after 160 seconds and prior to 308 seconds.  I'm not
sure how you concluded it took 182 seconds based on the log file I have.



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.