Author: Uri Blass
Date: 02:45:23 12/09/02
Go up one level in this thread
On December 08, 2002 at 14:37:16, Gian-Carlo Pascutto wrote: >On December 08, 2002 at 09:26:32, K. Burcham wrote: > >> >> >>have you made any changes since posting this? I dont know how old this version >>of your suite is or if you have made any changes since. > >There have been no changes. Uri found a position that had a second >solution, but I don't remember which one it was. > >I am probably going to update this someday, but no time for that >right now. Nevertheless the old version is still very good compared to >most other testsuites out there (IMHO). > >-- >GCP I think that the main problem with the test suite is that programs can find the right move for the wrong reason for example in the first position programs may find Bg4 from the first ply. more than one solution seems to be a problem in more than one position if you define every winning move as a solution. I decided to check the first position for more than one solution and it seems that d4 is also winning(d4 is considered to be second best move by programs). A possible computer line after 1...d4 is 2.Bf2 Be6 3.Rg1 Bd5 4.Qd1 Ng4 5.Bf5 Nxf2 6.Bxh3 Nxd1 7. rxd1 d3 8.Rg2 Bxg2 9.Bxg2 Re2+ 10.Kxd3 Rxg2 with almost +5 When I give yace to learn from that line it tries to improve by 7.Kxd3 -4.66/12 When I goto previous moves yace says that they are forced and only when I go back to 3.Rg1 yace gives me alternative of 3.Qxc6 but 3.Qxc6 is also bad because of 3...Bb4+. It is typical for programs to see Bg4 and not d4 for tactical reasons so practically programs that play d4 do not see the tactics. Uri
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.