Author: Jon Dart
Date: 21:02:37 11/23/99
Go up one level in this thread
On November 23, 1999 at 20:10:21, James Robertson wrote: >On November 23, 1999 at 20:02:09, Jon Dart wrote: > >>In position #1, Crafty 16.16 switches from Qxh4 to Qh6 at 4 ply >>in under 1 second. So does my program Arasan. The eval is around >>-1.8, but Qxh4 is evaluated much worse. >> >>--Jon > >My program does the same, but later it chooses Qxh4 again because the threat it >saw before was not real. It again switches back to Qh6 at ply 12, though. Do >Arasan or Crafty do the same? > No, neither seems to behave this way. I let Arasan go 12 ply and got this: ply 1. Qxh4 1 seconds. score: -0.77 2 nodes. ply 1. Qxh4 1 seconds. score: 0.30 39 nodes. ply 2. Qxh4 1 seconds. score: 0.30 265 nodes. ply 2. Qxh4 1 seconds. score: -0.33 306 nodes. ply 2. Qxh4 1 seconds. score: -0.33 344 nodes. ply 3. Qxh4 1 seconds. score: -0.42 2014 nodes. ply 4. Qxh4 1 seconds. score: -0.42 6541 nodes. ply 4. Qh6 1 seconds. score: -1.70 9663 nodes. ply 4. Qh6 1 seconds. score: -1.70 10435 nodes. ply 5. Qh6 1 seconds. score: -1.50 17549 nodes. ply 6. Qh6 1 seconds. score: -1.58 45751 nodes. ply 7. Qh6 2 seconds. score: -1.55 129520 nodes. ply 8. Qh6 6 seconds. score: -1.67 458897 nodes. ply 9. Qh6 15 seconds. score: -1.64 1195065 nodes. ply 10. Qh6 33 seconds. score: -1.77 2757884 nodes. ply 11. Qh6 77 seconds. score: -1.77 6875898 nodes. ply 12. Qh6 299 seconds. score: -1.67 26418042 nodes. Maybe you've got over-aggressive forward pruning or lazy eval? (It could be a lot of other things, but these come to mind). --Jon
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.