Author: Robert Hyatt
Date: 18:38:23 09/06/02
Go up one level in this thread
On September 06, 2002 at 21:07:47, scott farrell wrote:
>On September 06, 2002 at 16:10:56, Robert Hyatt wrote:
>
>>On September 06, 2002 at 05:17:01, David Dory wrote:
>>
>>>On September 05, 2002 at 21:46:41, scott farrell wrote:
>>>
>>>>Hi guys,
>>>>
>>>>I hope someone can give me some ideas, or can understand this pattern.
>>>>
>>>>this is the fen position
>>>>
>>>>[D]7k/pp3p1p/1np5/5N2/1P2P1NR/7P/3r2PK/2r5 b - - 0 50
>>>>
>>>>clearly the best move is Nd7 (aka B6D7)
>>>>Rcc2 (aka c1c2) is not too bad either
>>>>The rest pretty much lead to mate.
>>>>My program works that out pretty quickly, but changes its mind during depth 10
>>>>to D2C2.
>>>>When searching depth 11 - its sees that D2C2 leads straight to being mated (in 2
>>>>plies - mate in 1 - I think). The move ordering obviously gets screwed up, and
>>>>it does an almost reverse alpha-beta search.
>>>>I assume this is a bug somewhere in my search, does anyone have any ideas?
>>>>
>>>>this is my programs analysis:
>>>>
>>>>d6/8(3) 4.28 0.38secs 6820 C1C2 G4E3 C2A2 H4G4 B6D7 G4G7
>>>>->6/10(3) 4.28 0.71secs 38971 C1C2 G4E3 C2A2 H4G4 B6D7 G4G7
>>>>d7/12(3) 4.59 0.88secs 55850 C1C2 F5E3 C2B2 G4F6 H8G7 F6H7 B2B4
>>>>d7/12(4) 4.85 1.7secs 137123 B6D7 F5H6 F7F6 G4E3 H8G7 H6F5
>>>>->7/12(4) 4.85 1.7secs 137141 B6D7 F5H6 F7F6 G4E3 H8G7 H6F5
>>>>d8/12(3) 4.22 1.98secs 162918 B6D7 F5H6 F7F6 H4H5 C1C2 H6F7
>>>>->8/14(4) 4.22 4.01secs 364580 B6D7 F5H6 F7F6 H4H5 C1C2 H6F7
>>>>d9/16(4) 4.69 6.87secs 654405 B6D7 H4H6 C1C2 G4E3 C2B2 H6D6 D2D6
>>>>->9/16(4) 4.69 7.36secs 711939 B6D7 H4H6 C1C2 G4E3 C2B2 H6D6 D2D6
>>>>d10/18(4) 3.86 12.3secs 1292913 B6D7 H4H6 C6C5 B4B5 C1E1 F5D6 E1F1 D6C4 D2C2
>>>>H6D6
>>>>d10/18(4) 3.9 19.17secs 1941731 D2C2 H2G3 C6C5 B4C5
>>>>->10/24(4) 3.9 30.43secs 3164993 D2C2
>>>>d11/19(2) -501.99 30.59secs 3178776 PV miss at root:D2C2
>>>>d11/20(3) -1.03 41.52secs 4470207 D2G2 H2G2 C1C2 G2F3 C2C3 F3E2 B6D7 F5D6
>>>>H8G7 D6B7 C3G3 G4F2 G7F6
>>>>d11/20(3) 3.89 43.01secs 4611313 C1C2 F5E3 C2B2 G4F6 H8G7 E4E5 G7F8 F6H7
>>>>d11/22(4) 4.59 116.28secs 11658630 B6D7 G4E3 D7E5
>>>>
>>>>This is from crafty:
>>>>
>>>>({10:Mat06} 35. ... Ra2 36. Nf6 Rxg2+ 37. Kxg2 Rc2+ 38. Kg3 Rc3+ 39. Kg4 Rxh3
>>>>40. Rxh3 Nc4 41. Rxh7# +-)
>>>>(yes beleive it or not, my engine played Ra2 - a few months back without
>>>>extensions)
>>>>({10:-5.14} 35. ... Nd7 36. Rh6 Rb1 37. Rd6 Rxd6 38. Nxd6 Kg7 39. Nxb7 h5 40.
>>>>Ne3 Rxb4 -+)
>>>>({10:-1.74} 35. ... Rcc2 36. Nf6 Rxg2+ 37. Kh1 Rh2+ 38. Kg1 Rcg2+ 39. Kf1 Rg7
>>>>40. Nxg7 Kxg7 41. Ne8+ Kg8 42. Nd6 Rb2 43. Rf4 Rb1+ 44. Kg2 Rb2+ 45. Kf3 -+)
>>>>({10:+2.46} 35. ... Nd5 36. exd5 Rcc2 37. Nge3 Rb2 38. Re4 h5 39. dxc6 bxc6 40.
>>>>Ng3 c5 41. bxc5 +-)
>>>>
>>>>
>>>>normal moves:
>>>>C1H1,C1G1,C1F1,C1E1,C1D1,C1B1,C1A1,C1C2,C1C3,C1C4,C1C5,D2D1,D2G2,D2F2,D2E2,D2C2,
>>>>D2B2,D2A2,D2D3,D2D4,D2D5,D2D6,D2D7,D2D8,C6C5,B6C4,B6A4,B6D5,B6D7,B6C8,B6A8,H7H5,
>>>>H7H6,F7F6,A7A5,A7A6,H8G7,H8G8,
>>>>OUT OF CHECK moves:
>>>>C1H1,C1G1,C1F1,C1E1,C1D1,C1B1,C1A1,C1C2,C1C3,C1C4,C1C5,D2D1,D2G2,D2F2,D2E2,D2C2,
>>>>D2B2,D2A2,D2D3,D2D4,D2D5,D2D6,D2D7,D2D8,C6C5,B6C4,B6A4,B6D5,B6D7,B6C8,B6A8,H7H5,
>>>>H7H6,F7F6,A7A5,A7A6,H8G7,H8G8,
>>>>qsearch moves:
>>>>D2G2,
>>>>sideotmove token:
>>>>7k/pp3p1p/1np5/5N2/1P2P1NR/7P/3r2PK/2r5/ b - - 0 50
>>>>searchProbeThread starting
>>>>start search from 6 to 30
>>>>
>>>>Scott
>>
>>
>>I don't see this as a "pattern" problem. It looks like you need to check
>>two things:
>>
>>(1) your search extensions. IE are you extending on checks at least? and
>>a whole ply?
>>
>>(2) are you sure your checkmate detection is working? Allowing a short mate
>>with a 10 ply search suggests you are overlooking something. The question is,
>>what? and why?
>>
>>Are you doing any odd stuff like futility pruning, razoring, null-move, etc?
>
>Robert,
>
>No futility code - other than 'stand-pat' in the Qsearch.
>
>yes, adaptive null move.
>
>I still cant find it.
>
>Scott
OK... let's get specific.
Exactly how long does it take you to find the mate if _you_ play the
rook move yourself? I mean you play the move and let your program search
the resulting position here...
Then we can figure out if your extensions are working by seeing how
deep you have to go to find that mate from the white side after black
blows it...
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.