Author: Peter McKenzie
Date: 00:45:16 03/29/00
Go up one level in this thread
On March 28, 2000 at 17:13:08, Ernst A. Heinz wrote: >Hi Peter, Hi Ernst! > >>[D]3qkbr1/1b1n1p1p/4pp2/rB6/P2p1B2/5N2/1P3PPP/R2Q1RK1 b - - > >Your node explosion might just be an artifact of "LambChop" >for this special position. Things like this simply happen >(e.g. because of slightly unstable evaluation scores causing >unfortunate researches at interior nodes). That might be the case, but I'll reserve judgement until I can prove it one way or another. I will try to measure the stability of the search and or evaluation, not sure exactly how yet. Its hard to know what is going on in a search tree sometimes. Do you ever see DarkThought require a branching factor of 10-15 for occasional positions? > >"DarkThought WCCC'99" does unlimited recapture extensions >(in Hitech style, i.e., capture bringing score close back to >root value) with a full ply increase. Yet, its effective >branching factor remains smaller than 2.5 in iteration #12. > >00:03:08 11.01 e5 Bg3 Be7 Nh4 Ba6 Bxa6 Rxa6 Qd3 Rb6 Qxh7 =-0.88 #61077398 >00:07:22 12.01 e5 Bg3 Be7 Nh4 Ba6 Bxa6 Rxa6 Qd3 Nc5 Qxh7 =-0.95 #142815058 > >That's why I like to speak of "scalable search" in connection >with "DarkThought" ... :-) > >=Ernst= > >P.S. >WWW pages of "DarkThought" at URL >http://supertech.lcs.mit.edu/~heinz/dt/.
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.