Author: George Tsavdaris
Date: 11:08:52 01/03/06
Conditions: A match between Fruit2.1 Vs BrFruit 2.1 with depth = D for Fruit 2.1 and D+1 for BrFruit2.1 for D=1,2,3,....,13 (I haven't yet managed to complete the D=13 but i will do it soon.....) -Fruit 2.1 is not using it's own book, and uses default parameters and it is playing with depth D (that means it moves when it completes D ply). -BrFruit 2.1 is Fruit 2.1 renamed(so i could run this match to DeepFritz8 GUI) with default parameters and no book also and it is playing with depth D+1 (that means it moves when it completes D+1 ply). For each D there is a match of 10 games between these 2 engines, from a set of 5 standard positions. Each engine will play these 5 positions with black and again the same 5 with white.... So 10 games will occur for each D (In the future, i will also add to these results, games that they will play with another set of 5 positions). Positions: 1)Sicilian Najdorf, 6.Be2 r2q1rk1/1p1nbppp/p2pbn2/4p3/4P3/1NN1B3/PPPQBPPP/R4RK1 w - - 0 11 2)Sicilian Najdorf, Poisoned Pawn. rnb1kb1r/1p3ppp/pq1ppn2/6B1/3NPP2/2N5/PPP3PP/R2QKB1R w KQkq - 0 8 3)Closed Ruy Lopez, Chigorin Defence. r1b2rk1/2q1bppp/p2p1n2/np2p3/3PP3/5N1P/PPBN1PP1/R1BQR1K1 b - - 0 13 4)Sicilian Dragon, Yugoslav Attack, 9.Bc4 2rq1rk1/pp1bppb1/3p1np1/4n2p/3NP2P/1BN1BP2/PPPQ2P1/2KR3R w - h6 0 13 5)Queen's Indian, 4.g3 r2q1rk1/p2nbppp/bpp1p3/3p4/2PP4/1PB3P1/P2NPPBP/R2Q1RK1 b - - 0 12 The winning percentage of the BrFruit 2.1 that was looking one ply higher, was almost the same for D=1,2,3,...,11 but at D=12 the strength of 2 engines appeared more close than on other D-D+1 plies. I can't be sure for anything until i add at least 20 more games(from 10 more predefined positions) for each D and until i run D=13 or more also....... A graph of (% percentage of BrFruit 2.1) - (D) can be seen from: http://www.angelfire.com/ultra/georgecm9000/alt0189.html How the graph should be from a theoretical point of view? Should it have the form Percentage(D) = a^(-b·D) + 0.5 (a>0,b>0) or it should be another way? All games can be downloaded from: http://www.angelfire.com/ultra/georgecm9000/alt0189.html Results: 1Ply / 2Ply 1234567890 BrFruit 2.1 111111101½ 8.5/10 Fruit 2.1 000000010½ 1.5/10 2Ply / 3Ply 1234567890 BrFruit 2.1 11111½1101 8.5/10 Fruit 2.1 00000½0010 1.5/10 3Ply / 4Ply 1234567890 BrFruit 2.1 1½11½11101 8.0/10 Fruit 2.1 0½00½00010 2.0/10 4Ply / 5Ply 1234567890 BrFruit 2.1 10110½1½½1 6.5/10 Fruit 2.1 01001½0½½0 3.5/10 5Ply / 6Ply 1234567890 BrFruit 2.1 01½11½1011 7.0/10 Fruit 2.1 10½00½0100 3.0/10 6Ply / 7Ply 1234567890 BrFruit 2.1 ½11½111101 8.0/10 Fruit 2.1 ½00½000010 2.0/10 7Ply / 8Ply 1234567890 BrFruit 2.1 1110½½11½1 7.5/10 Fruit 2.1 0001½½00½0 2.5/10 8Ply / 9Ply 1234567890 BrFruit 2.1 1½11½11011 8.0/10 Fruit 2.1 0½00½00100 2.0/10 9Ply / 10Ply 1234567890 BrFruit 2.1 10½½11½½11 7.0/10 Fruit 2.1 01½½00½½00 3.0/10 10Ply / 11Ply 1234567890 BrFruit 2.1 11110½½0½½ 6.0/10 Fruit 2.1 00001½½1½½ 4.0/10 11Ply / 12Ply 1234567890 BrFruit 2.1 ½½11½½½1½1 7.0/10 Fruit 2.1 ½½00½½½0½0 3.0/10 12Ply / 13Ply 1234567890 BrFruit 2.1 1½0½½½1½½½ 5.5/10 Fruit 2.1 0½1½½½0½½½ 4.5/10 Draw percentage increases as D increases, i can't clearly explain why.....?!?!
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.