Progress 10 10 10 Personality is very strong

Discussion of computer chess matches and engine tournaments.

Moderators: hgm, Dann Corbit, Harvey Williamson

Forum rules
This textbox is used to restore diagrams posted with the [d] tag before the upgrade.
Post Reply
pichy
Posts: 2564
Joined: Thu Mar 09, 2006 2:04 am

Progress 10 10 10 Personality is very strong

Post by pichy » Thu Apr 26, 2007 1:07 am

The Movei00_8_403 with progress 10 10 10 Personality and Movei00_8_418 with the standard progress personality setup for comparison.

PS: So far after 14 games the Progress 10 10 10 Personality is performing better than the newer version of Movei 418 :shock:


Engine Score Mo Al S-B
1: Movei00_8_403 7.0/14 0010=111=01001 49.00
1: AlaricWB704 7.0/14 1101=000=10110 49.00

14 of 30 games played
Name of the tournament: Arena tournament
Site/ Country: Jorge, United States
Level: Blitz 5/1
Hardware: AMD Athlon(tm) Processor 1202 MHz with 512MB Memory
Operating system: Microsoft Windows XP Professional Service Pack 2 (Build 2600)
PGN-File: C:\Program Files\Arena\Arena.pgn
Website:
E-Mail Address:

______________________________________________________________________

Engine Score Al Mo S-B
1: AlaricWB704 18.5/30 1011100101111=0=1110==1===0101 212.75
2: Movei00_8_418 11.5/30 0100011010000=1=0001==0===1010 212.75

30 games played / Tournament finished
Name of the tournament: Arena tournament
Site/ Country: Jorge, United States
Level: Blitz 5/1
Hardware: AMD Athlon(tm) Processor 1202 MHz with 512MB Memory
Operating system: Microsoft Windows XP Professional Service Pack 2 (Build 2600)
PGN-File: C:\Program Files\Arena\Arena.pgn
Website:
E-Mail Address:

pichy
Posts: 2564
Joined: Thu Mar 09, 2006 2:04 am

Re: Alaric 704 Should have NOT resigned on this position....

Post by pichy » Thu Apr 26, 2007 4:14 am

pichy wrote:The Movei00_8_403 with progress 10 10 10 Personality and Movei00_8_418 with the standard progress personality setup for comparison.
pichy wrote:The Movei00_8_403 with progress 10 10 10 Personality and Movei00_8_418 with the standard progress personality setup for comparison.

Here is the most strange game where Alaric 704 resigned against Movei with only a King and Bishop, the game should had been a draw, this is the final position :shock:

PS: White should have NOT resigned against Black King and Bishop, Black can NOT mate, only win on time.


[D]8/8/8/8/b7/P7/2k5/6K1 w - - 0 132
[Event "Arena tournament"]
[Site "Jorge"]
[Date "2007.04.25"]
[Round "30"]
[White "AlaricWB704"]
[Black "Movei00_8_403"]
[Result "0-1"]
[ECO "A13"]
[Opening "English"]
[Time "23:36:10"]
[Variation "1...e6 2.Nc3 d5"]
[TimeControl "300+1"]
[Termination "normal"]
[PlyCount "261"]
[WhiteType "program"]
[BlackType "program"]

1.c4 e6 2.Nc3 d5 3.g3 d4 4.Nb5 e5 5.Bg2 c5 6.d3 Nc6 7.Nf3 Be7 8.O-O Nf6 9.e4 O-O
10.Nh4 a6 11.Na3 g6 12.f4 exf4 13.Bxf4 Ng4 14.Nc2 Bxh4 15.gxh4 f6 16.Qd2 Nge5 17.Rf2
b5 18.b3 bxc4 19.bxc4 Rb8 20.Ne1 Bg4 21.h3 Be6 22.Nf3 Rb4 23.Raf1 Rf7 24.Nxe5 Nxe5
25.Bg3 Nd7 26.h5 g5 27.Kh2 h6 28.Bf3 Qb6 29.Qe2 Rb1 30.Bg4 Rxf1 31.Qxf1 Rf8 32.Be2 Kg7
33.h4 Re8 34.hxg5 hxg5 35.Kh1 Rh8 36.Rh2 Qb4 37.Bf3 Qa3 38.Bd6 Rh6 39.Qd1 Qb4 40.Rg2
Qb6 41.Bh2 Qd8 42.e5 Nxe5 43.Bxe5 fxe5 44.Qd2 g4 45.Bxg4 Qh4+ 46.Rh2 Qxg4 47.Rg2
Rxh5+ 48.Kg1 Rg5 49.Rxg4 Rxg4+ 50.Kf2 Bf5 51.Qe1 e4 52.dxe4 Bxe4 53.Qa5 Rg2+ 54.Kf1
d3 55.Qxc5 Kf6 56.Qd6+ Kf5 57.Qf8+ Ke5 58.Qh8+ Ke6 59.Qc8+ Ke5 60.Qc7+ Kd4 61.Qd6+
Kxc4 62.Qxa6+ Kc5 63.Qxd3 Bxd3+ 64.Kxg2 Bc4 65.Kg1 Kb4 66.Kf2 Ka3 67.Kg1 Bb5 68.Kh1
Kb4 69.Kg2 Bc6+ 70.Kg1 Ba4 71.Kf2 Kb5 72.Kg1 Kc4 73.Kf2 Kd3 74.Kg1 Kd2 75.Kf2 Kc1
76.Ke3 Kb2 77.Kf2 Ka3 78.Ke2 Bb5+ 79.Kf2 Bd7 80.Ke1 Bc8 81.Kf2 Ba6 82.Kg1 Bc4 83.Kf2
Bf7 84.Kg1 Bg6 85.Kf2 Be4 86.Kg1 Ka4 87.Kf2 Bb7 88.Ke2 Bg2 89.Kf2 Ba8 90.Kg1 Ka3
91.Kf2 Bb7 92.Ke2 Bh1 93.Kf2 Ba8 94.Kg1 Be4 95.Kf2 Bh1 96.Ke2 Bg2 97.Kf2 Bd5 98.Kg1
Kb4 99.Kf2 Bg8 100.Kg1 Kc5 101.a3 Be6 102.Kf2 Kb5 103.Kg1 Ka4 104.Kf2 Bd5 105.Kg1
Bf7 106.Kf2 Bg8 107.Kg1 Bd5 108.Kf2 Be6 109.Kg1 Ba2 110.Kf2 Ka5 111.Ke1 Kb6 112.Kf2
Kb7 113.Ke1 Ka6 114.Kf2 Bg8 115.Ke1 Kb6 116.Kf2 Bb3 117.Kg1 Ba4 118.Kf2 Kb5 119.Kg1
Kc4 120.Kf2 Kd3 121.Kg1 Ke2 122.Kh2 Kf2 123.Kh1 Kf1 124.Kh2 Ke2 125.Kg1 Kf3 126.Kh2
Ke3 127.Kg1 Kd2 128.Kh2 Kc3 129.Kg1 Kb2 130.Kf2 Kc2 131.Kg1 {White resigns} 0-1





_________________________________________________


PS: Both Program versions Movei 8_403 Progress 10 10 10 Personality and Movei 8_418 ended with the same score :shock:

Engine Score Al Mo S-B
1: AlaricWB704 18.5/30 1011100101111=0=1110==1===0101 212.75
2: Movei00_8_418 11.5/30 0100011010000=1=0001==0===1010 212.75

30 games played / Tournament finished
Name of the tournament: Arena tournament
Site/ Country: Jorge, United States
Level: Blitz 5/1
Hardware: AMD Athlon(tm) Processor 1202 MHz with 512MB Memory
Operating system: Microsoft Windows XP Professional Service Pack 2 (Build 2600)
PGN-File: C:\Program Files\Arena\Arena.pgn
Website:
E-Mail Address:
_____________________________________________________

Engine Score


Al Mo S-B
1: AlaricWB704 18.5/30 1101=000=101101=1==111=110=110 212.75
2: Movei00_8_403 11.5/30 010=111=010010=0==000=001=001 212.75

30 games played / Tournament finished
Name of the tournament: Arena tournament
Site/ Country: Jorge, United States
Level: Blitz 5/1
Hardware: AMD Athlon(tm) Processor 1202 MHz with 512MB Memory
Operating system: Microsoft Windows XP Professional Service Pack 2 (Build 2600)
PGN-File: C:\Program Files\Arena\Arena.pgn
Website:
E-Mail Address:

Uri Blass
Posts: 8927
Joined: Wed Mar 08, 2006 11:37 pm
Location: Tel-Aviv Israel

Re: Progress 10 10 10 Personality is very strong

Post by Uri Blass » Thu Apr 26, 2007 4:41 am

My testing so far suggest that 403 progress 10 10 10 may be better than 418 progress 10 10 10 but it is possible that 403 was simply lucky.

403 progress 10 10 10 was the only program that could beat fruit2.0 in 100 game match 51-49 in 5 minutes/40 moves.

418 progress 10 10 10 will probably lose because after 62 games the result is 32.5-29.5 for fruit2.0

result of 403 progress 10 10 10(in the match that it won 51-49) after the same number of games was 32.5-29.5 for movei

Of course it proves nothing and this difference may be a statistical error.

Uri

Peter Fendrich

Re: Alaric 704 Should have NOT resigned on this position....

Post by Peter Fendrich » Thu Apr 26, 2007 8:04 am

Thank you for this position!

You should turn off resign in Alaric.ini
The resign function is buggy and will resign in some cases where there is a dead draw.

/Peter

pichy
Posts: 2564
Joined: Thu Mar 09, 2006 2:04 am

Re: Movei Progress 10 10 10 vs Fruit ended 20.5 to 19.5

Post by pichy » Thu Apr 26, 2007 12:56 pm

Uri Blass wrote:My testing so far suggest that 403 progress 10 10 10 may be better than 418 progress 10 10 10 but it is possible that 403 was simply lucky.

403 progress 10 10 10 was the only program that could beat fruit2.0 in 100 game match 51-49 in 5 minutes/40 moves.

418 progress 10 10 10 will probably lose because after 62 games the result is 32.5-29.5 for fruit2.0

result of 403 progress 10 10 10(in the match that it won 51-49) after the same number of games was 32.5-29.5 for movei

Of course it proves nothing and this difference may be a statistical error.

Uri

This was a very close result, Fruit 2.0 was able to catch up to Movei 8_403 in the last 12 games :shock:


Engine Score Mo Fr S-B
1: Movei00_8_403 20.5/40 00=111=11011001110=1=1101==100000==0==0= 399.75
2: Fruit20 19.5/40 11=000=00100110001=0=0010==011111==1==1= 399.75

40 games played / Tournament finished
Name of the tournament: Arena tournament
Site/ Country: Jorge, United States
Level: Blitz 5/1
Hardware: AMD Athlon(tm) Processor 1202 MHz with 512MB Memory
Operating system: Microsoft Windows XP Professional Service Pack 2 (Build 2600)
PGN-File: C:\Program Files\Arena\Arena.pgn
Website:
E-Mail Address:

pichy
Posts: 2564
Joined: Thu Mar 09, 2006 2:04 am

Re: Progress 10 10 10 Personality is very strong

Post by pichy » Sun May 06, 2007 1:11 am

I have done several testings with different progresses, and finally after more than 500 games at time control of 5/1 the best result are with progress 8 8 8 which is turning better results than with progress 10 10 10 :shock:

Uri Blass wrote:My testing so far suggest that 403 progress 10 10 10 may be better than 418 progress 10 10 10 but it is possible that 403 was simply lucky.

403 progress 10 10 10 was the only program that could beat fruit2.0 in 100 game match 51-49 in 5 minutes/40 moves.

418 progress 10 10 10 will probably lose because after 62 games the result is 32.5-29.5 for fruit2.0

result of 403 progress 10 10 10(in the match that it won 51-49) after the same number of games was 32.5-29.5 for movei

Of course it proves nothing and this difference may be a statistical error.

Uri

Uri Blass
Posts: 8927
Joined: Wed Mar 08, 2006 11:37 pm
Location: Tel-Aviv Israel

Re: Progress 10 10 10 Personality is very strong

Post by Uri Blass » Sun May 06, 2007 6:02 am

pichy wrote:I have done several testings with different progresses, and finally after more than 500 games at time control of 5/1 the best result are with progress 8 8 8 which is turning better results than with progress 10 10 10 :shock:

Uri Blass wrote:My testing so far suggest that 403 progress 10 10 10 may be better than 418 progress 10 10 10 but it is possible that 403 was simply lucky.

403 progress 10 10 10 was the only program that could beat fruit2.0 in 100 game match 51-49 in 5 minutes/40 moves.

418 progress 10 10 10 will probably lose because after 62 games the result is 32.5-29.5 for fruit2.0

result of 403 progress 10 10 10(in the match that it won 51-49) after the same number of games was 32.5-29.5 for movei

Of course it proves nothing and this difference may be a statistical error.

Uri
You may be correct

I did not test 8 8 8

I am going to test 8 8 8 to see if it performs better than 10 10 10.

result of 10 10 10 against other engines are the following:

5 minutes/40 moves:beating fruit2.0 51-49
3 minutes/40 moves FRC:losing against glaurung1.2.1 32-18

pichy
Posts: 2564
Joined: Thu Mar 09, 2006 2:04 am

Re: Progress 10 10 10 Personality is very strong

Post by pichy » Sun May 06, 2007 10:11 am

Uri Blass wrote:
pichy wrote:I have done several testings with different progresses, and finally after more than 500 games at time control of 5/1 the best result are with progress 8 8 8 which is turning better results than with progress 10 10 10 :shock:

Uri Blass wrote:My testing so far suggest that 403 progress 10 10 10 may be better than 418 progress 10 10 10 but it is possible that 403 was simply lucky.

403 progress 10 10 10 was the only program that could beat fruit2.0 in 100 game match 51-49 in 5 minutes/40 moves.

418 progress 10 10 10 will probably lose because after 62 games the result is 32.5-29.5 for fruit2.0

result of 403 progress 10 10 10(in the match that it won 51-49) after the same number of games was 32.5-29.5 for movei

Of course it proves nothing and this difference may be a statistical error.

Uri
You may be correct

I did not test 8 8 8

I am going to test 8 8 8 to see if it performs better than 10 10 10.

result of 10 10 10 against other engines are the following:

5 minutes/40 moves:beating fruit2.0 51-49
3 minutes/40 moves FRC:losing against glaurung1.2.1 32-18
Sorry Uri, but the progress the performed the best was NOT 8 8 8 that was the previous progress that I tested before progress 11 11 11. Here is the configuration Settings for Movei_changes403:


progress 11 11 11 0
loghash 20 loghasheval 21

Uri Blass
Posts: 8927
Joined: Wed Mar 08, 2006 11:37 pm
Location: Tel-Aviv Israel

Re: Progress 10 10 10 Personality is very strong

Post by Uri Blass » Sun May 06, 2007 10:39 am

You may be correct.
I stopped the match of glaurung against progress 8 8 8 after glaurung was leading 17.5-6.5(only 27% for movei 8 8 8 when movei 10 10 10 achieved 36% after 50 games against glaurung)

unfortunately I limited the progress parameter to 10 in the uci options
because I did not believe that big values are going to work.

I expect the difference between 11 and 10 to be small.
If you want to get a big change(big improvement or big reduction in playing strength) then you may try 20 10 10 or 20 20 20.

note that I never tried it.

Uri

pichy
Posts: 2564
Joined: Thu Mar 09, 2006 2:04 am

Re: Progress 10 10 10 Personality is very strong

Post by pichy » Mon May 07, 2007 9:27 pm

Uri Blass wrote:
pichy wrote:I have done several testings with different progresses, and finally after more than 500 games at time control of 5/1 the best result are with progress 8 8 8 which is turning better results than with progress 10 10 10 :shock:

Uri Blass wrote:My testing so far suggest that 403 progress 10 10 10 may be better than 418 progress 10 10 10 but it is possible that 403 was simply lucky.

403 progress 10 10 10 was the only program that could beat fruit2.0 in 100 game match 51-49 in 5 minutes/40 moves.


Of course it proves nothing and this difference may be a statistical error.

Uri you are correct to say that it proves nothing since my result is very bad for movei 403 progress 10 10 10 against Fruit 2.0 :(

Uri
result of 10 10 10 against other engines are the following:

5 minutes/40 moves:beating fruit2.0 51-49


I just stoped a match at game 60 with progress 10 10 10 since movei403 was losing to fruit, it won't be necessary to continue the match since I don't think that movei403 will make a huge comeback.


Engine Score Fr Mo S-B
1: Fruit20 35.5/60 ···························································· 10=0=1=10101==01==0=11111110010==1011=01=11101101=0==0=11=== 869.75
2: Movei00_8_403 24.5/60 01=1=0=01010==10==1=00000001101==0100=10=00010010=1==1=00=== ···························································· 869.75

60 of 100 games played
Name of the tournament: Arena tournament
Site/ Country: Jorge, United States
Level: Blitz 5/1
Hardware: AMD Athlon(tm) Processor 1202 MHz with 512MB Memory
Operating system: Microsoft Windows XP Professional Service Pack 2 (Build 2600)
PGN-File: C:\Program Files\Arena\Arena.pgn
Website:
E-Mail Address:

Post Reply