Author: Robert Hyatt
Date: 18:31:24 01/10/00
Go up one level in this thread
On January 10, 2000 at 17:45:21, Jeremiah Penery wrote: >On January 10, 2000 at 16:30:58, Robert Hyatt wrote: > >>No program _ever_ found the draw. It is over 60 plies deep from the ending >>position... which is hopeless unless you start to give a program 24 hours or >>more to search. > >Just since this has come up... >I let my modified Crafty run on this for about a month over Xmas break...This is >what it came up with: > >I gave it the position before the final move of the game, and told it to search >that move. > >White(45): search Ra6 > time surplus 0.00 time limit 30.00 (3:00) > nss depth time score variation (1) > 1-> 0.01 -0.58 45. Ra6 Qxc6 46. dxc6 > 2-> 0.04 -0.58 45. Ra6 Qxc6 46. dxc6 > 3-> 0.06 -0.87 45. Ra6 Qxc6 46. dxc6 Bc7 > 4 0.07 ++ 45. Ra6!! > 4-> 0.09 0.33 45. Ra6 Qxc6 46. dxc6 Bc7 47. Ke2 > 5 0.10 -- 45. Ra6 > 5-> 0.14 -0.07 45. Ra6 Qxc6 46. dxc6 Bc7 47. Ke2 Ke7 > 6 0.17 ++ 45. Ra6!! > 6-> 0.19 0.32 45. Ra6 Qxc6 46. dxc6 Bc7 47. Ke2 Ke7 > 7 0.28 ++ 45. Ra6!! > 7-> 0.41 0.89 45. Ra6 Qxc6 46. dxc6 h5 47. Ra7+ Be7 > 48. Ke2 Rd8 49. Ke3 > 8 0.61 ++ 45. Ra6!! > 8-> 0.90 1.28 45. Ra6 Qxc6 46. dxc6 h5 47. Ra7+ Be7 > 48. Ke2 Rd8 49. Ke3 > 9-> 1.40 1.22 45. Ra6 Qxc6 46. dxc6 Ke8 47. Ra7 Rd8 > 48. Rb7 Ra8 49. Rxb5 Bc7 > 10 3.44 ++ 45. Ra6!! > 10-> 3.76 1.62 45. Ra6 Qxc6 46. dxc6 Ke8 47. Ra7 Rc8 > 48. Ke2 Rc7 49. Ra5 Ke7 50. Rxb5 > 11-> 5.78 1.61 45. Ra6 Qxc6 46. dxc6 Kf8 47. Ra7 Rc8 > 48. Ke2 h5 49. Rb7 Rc7 50. Rxb5 Ke7 > 12-> 20.51 1.72 45. Ra6 Qxc6 46. dxc6 Kf8 47. Ra7 Rc8 > 48. Ke2 Bc7 49. Rb7 Ke7 50. Rxb5 Kd6 > 51. Ke3 > 13-> 2:10 1.68 45. Ra6 Qe3 46. Qd7+ Kg8 47. Qxd6 Rf8 > 48. Qe6+ Kh8 49. Qe7 Rg8 50. Bf3 Qc1+ > 51. Kf2 Qd2+ 52. Be2 Qf4+ 53. Ke1 Qc1+ > 54. Kf2 Qxc3 > 14 3:26 -- 45. Ra6 > 14-> 7:46 0.51 45. Ra6 Qe3 46. Qxd6 Re8 47. h4 h5 > 48. Bf3 Qc1+ 49. Kf2 Qd2+ 50. Be2 Qf4+ > 51. Kg1 Qe3+ 52. Kh2 Qf4+ 53. Kh3 Qxf5+ > 54. Kh2 Qh7 > 15-> 12:51 0.44 45. Ra6 Qe3 46. Qxd6 Re8 47. h4 h5 > 48. Bf3 Qc1+ 49. Kf2 Qd2+ 50. Be2 Qf4+ > 51. Kg1 Qe3+ 52. Kh2 Qf4+ 53. g3 Qf2+ > 54. Kh3 Qxe2 55. Qd7+ Kf8 56. Qxb5 > 16-> 24:33 0.65 45. Ra6 Qe3 46. Qxd6 Re8 47. h4 h5 > 48. Bf3 Qc1+ 49. Kf2 Qd2+ 50. Be2 Qf4+ > 51. Kg1 Qe3+ 52. Kh2 Qf4+ 53. Kh3 Qxf5+ > 54. Kh2 Qf4+ 55. Kg1 Qc1+ 56. Kf2 Qh6 > 17-> 62:44 0.65 45. Ra6 Qe3 46. Qxd6 Re8 47. h4 h5 > 48. Bf3 Qc1+ 49. Kf2 Qd2+ 50. Be2 Qf4+ > 51. Kg1 Qe3+ 52. Kh2 Qf4+ 53. Kh3 Qxf5+ > 54. Kh2 Qf4+ 55. Kg1 Qe3+ 56. Kf1 Qf4+ > 57. Ke1 <HT> > 18-> 257:24 0.73 45. Ra6 Qe3 46. Qxd6 Re8 47. h4 h5 > 48. Bf3 Qc1+ 49. Kf2 Qd2+ 50. Be2 Qf4+ > 51. Kg1 Qe3+ 52. Kh2 Qf4+ 53. Kh3 Qxf5+ > 54. Kh2 Qf4+ 55. Kg1 Qe3+ 56. Kf1 Qc1+ > 57. Kf2 Qf4+ 58. Bf3 Qh6 > 19-> 784:39 0.75 45. Ra6 Qe3 46. Qxd6 Re8 47. h4 h5 > 48. Bf3 Qc1+ 49. Kf2 Qd2+ 50. Be2 Qf4+ > 51. Kg1 Qe3+ 52. Kh2 Qf4+ 53. Kh3 Qxf5+ > 54. Kh2 Qf4+ 55. Kg1 Qe3+ 56. Kf1 Qc1+ > 57. Kf2 Qf4+ 58. Bf3 Qd2+ 59. Kf1 Qh6 > 20 5177:33 -- 45. Ra6 > >I was optimistic here...Up until now, it had been following the PV given on the >Rebel site as the longest drawing PV (36 ply - given below), and I hoped that it >might have discovered the quiet move (60. ...Qc1) in that line which would allow >the draw...It was not so, and it changed the entire PV. I did not know this >until 3 weeks later, however, because I left for break before the resolution of >the fail-low. > >-- >This is from Ed's page: >45.Ra6? Qe3! 46.Qxd6 Re8! 47.h4! h5! 48.Bf3 Qc1+ 49.Kf2 Qd2+ 50.Be2 Qf4+ 51.Kg1 >Qe3+ 52.Kh2 Qf4+ 53.Kh3 Qxf5+ 54.Kh2 Qf4+ 55.Kg1 Qe3+ 56.Kf1 Qc1+ 57.Kf2 Qf4+ >58.Ke1 Qc1+ 59.Bd1 Qxc3+ 60.Kf1 Qc1! 61.Ke2 Qb2+ 62.Kf1 Qc1 >-- > 20-> 8855:29 0.00 45. Ra6 Qe3 46. Qxd6 Re8 47. Qe6+ Rxe6 > 48. fxe6+ Ke7 49. Bg6 Qc1+ 50. Kf2 > Qb2+ 51. Kg3 Qxc3+ 52. Kh2 Qe3 53. > Bf5 Qf4+ 54. Kh1 Qxf5 55. Ra7+ Kd8 > 56. Ra8+ Kc7 57. e7 Qf1+ 58. Kh2 Qf4+ > 59. Kh1 Qc1+ 60. Kh2 Qf4+ > 21-> 20053:38 0.15 45. Ra6 Qe3 46. Qxd6 Re8 47. Qd7+ Re7 > 48. Qc6 Qxe4 49. d6 Qd3+ 50. Kg1 Re8 > 51. Ra7+ Kf8 52. Qc5 Rd8 53. Rxg7 Qd1+ > 54. Kh2 Qxd6 55. Rh7 Qxc5 56. bxc5 > Rd3 57. Rxh6 Ke7 > >I'm only left wondering why it left the PV it had before... > > 22 23388:31 ++ 45. Ra6!! > >Now I was disappointed that the score is going back up. > >It ended up searching about 40000 minutes before I stopped it - the fail-high >had not resolved. Total nodes ~= 315 billion. This may have been a bug. I had a minor problem with the dynamic drawscore stuff. ie normally draw=0, but can drop to -20, etc. And on a rare occasion here and there, it could be +20 due to a bug. I have recently discovered/fixed this. I let Crafty run on some powerful hardware after that match, and by playing the first couple of moves (Qe3 to start) it found a draw score after about a day and a half or so... But nobody found it in reasonable time...
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.