The FREEWARE Group Championship Matches Have Begun

[geots]

Group championship matches have begun running with:

Group 14:
----------
Marvin 1.3.0 v Smash 1.0.3

Group 13:
----------
KnockOut 0.7.1 v Parrot 07.07.22

Group 12:
----------
Capivara LK 0.08b03 64-bit v Shallow 1.0 64-bit

Group 11:
----------
Chispa 4.0.3 v Chezzz 1.0.3



As I stated earlier, these are 24 game matches. Only difference being I am doing something I have not done before. Instead of engine-engine matches, I am running each as a 24-cycle RR. This way stoppages (voluntary or otherwise) don't cause a split in the PGN files- for one thing. And all controls are exactly as they were for the RRs. Updates will be posted, but not after every move.



george

[geots]

These updates are for Groups 11 thru 14. The only problem- there is no update for Group 12. At present I have not been able to get them to run in a 24-cycle 2 engine RR. Both are causing problems and don't want to load for the next game when the previous one ends. I have given them enough of a chance that I will have to go to an engine-engine format. I assume that will solve the problem, as neither caused a problem in the original 14 engine RR. At any rate, I will do the best I can with them. To the ones running ok:

Code: Select all

Group 14 XPS-Core-i5  2013

smash 1.0.3     +6/-6/=5   50.00%   8.5/17   
Marvin 1.3.0    +6/-6/=5   50.00%   8.5/17 

Dead even after 17 games of the 24-game match.

Code: Select all

Group 13 XPS-Core-i5     2013

KnockOut 0.7.1     +8/-4/=6   61.11%   11.0/18
Parrot 07.07.22    +4/-8/=6   38.89%    7.0/18

This one I will make an "unofficial call" on. I don't believe Parrot can make up 4 points in 6 games. KnockOut will be the Group 13 champion. Where are you, Sven?!

Code: Select all

Group 11 SP8-Core-i5  2013

                  
Chezzz 1.0.3    +7/-3/=3   65.38%   8.5/13
Chispa 4.0.3    +3/-7/=3   34.62%   4.5/13

Also a 4 game lead here for Chezzz, but with 11 games still to play- it may or may not be safe.



We shall see-

george

[Sven]

Code: Select all

Group 13 XPS-Core-i5     2013

KnockOut 0.7.1     +8/-4/=6   61.11%   11.0/18
Parrot 07.07.22    +4/-8/=6   38.89%    7.0/18

This one I will make an "unofficial call" on. I don't believe Parrot can make up 4 points in 6 games. KnockOut will be the Group 13 champion. Where are you, Sven?!

I am here

Honestly, I did not even expect that KnockOut would participate in any recent tournament since I did not release any updates during the past 2,5 years (reason being that I stopped its development and work on something new). So many thanks for including KnockOut in your tourney!

As to the match results, well, I think that beating any of the engines from the groups above 13 will be quite tough for KnockOut, therefore I would not expect any miracles.

EDIT: Of course the match against Parrot is not decided yet, although the probability for Parrot to get at least 5 points in 6 games is around 4% only, assuming 33% draws and equal strength.

Sven

[geots]

Here are the final results for Groups 14, 13 and 11. Along with an update for Group 12 which couldn't decide if it wanted to play or not and had to be switched to engine-engine- thus a late start.




Inspiron 620 Intel i5 (SP8)-4 True Cores [40/21]
XPS 8500 Intel i5-4 True Cores [40/19]

Fritz 11 gui/Fritz 13 gui
1CPU/64-bit where available
128MB hash
Bases=NONE
Ponder_Learning=OFF
Perfect 2012b.ctg w/12-move limit
40/21 Repeating OR
40/19 Repeating (Benched to adapt to 40/40)
24-Game Matches

Code: Select all

Group 14 XPS-Core-i5  2013

smash 1.0.3    +29    +10/-8/=6   54.17%   13.0/24
Marvin 1.3.0   -29    +8/-10/=6   45.83%   11.0/24

Code: Select all

Group 13 XPS-Core-i5     2013

KnockOut 0.7.1    +58    +10/-6/=8   58.33%   14.0/24
Parrot 07.07.22   -58    +6/-10/=8   41.67%   10.0/24

Code: Select all

Group 11 SP8-Core-i5  2013

Chezzz 1.0.3   +104    +13/-6/=5   64.58%   15.5/24
Chispa 4.0.3   -104    +6/-13/=5   35.42%    8.5/24

And here is an update for Group 12, after 16 of 24 games:

Code: Select all

Capivara LK 0.08b03 64-bit     +9/-4/=3
Shallow 1.0 64-bit             +4/-9/=3

Again, another "unofficial call". Shallow cannot make up 5 points in 8 games. Capivara is too strong for him.




Group Champions:


Group 14- smash 1.0.3

Group 13- KnockOut 0.7.1 (Thanks for the reply, Sven. KnockOut records a "knockout".)

Group 12- Capivara LK 0.08b03 64-bit

Group 11- Chezzz 1.0.3



A special congratulations to the 4 winners. And to the runners-up as well. They all beat a hell of a lot of engines to get here. If there is anything sort of sad- it is that these engines are not all still in active development.

Groups 7 thru 10 will start their matches tomorrow. Along with an "official" final for Group 12.


Nite-

george

[geots]

As all the games from all the RRs are available in PGN format from the links given, I seriously doubt anyone would gain anything from the PGNs of these final matches. However, if requested, I will upload them and give you a link.


george

[geots]

I debated with myself as to whether posting this at this particular point was wise or not. But I need to be on record as having said it.

The champions of Groups 11 thru 14 have just as much right to be proud of what they have accomplished as do the champions of Groups 1 thru 4. They played to their level of competition, and fought, battled and beat some damn good engines. They should, and do, get just as much respect from me as the champions in the top groups.

Think of it this way, in a boxing analogy. All the fame and glory that Sugar Ray Leonard, Marvin Hagler, Roberto Duran and Chavez had- I can promise they never once felt short-changed because they knew they couldn't beat Ali for the heavyweight championship.

Be proud of what you have done here, and I consider myself lucky to have been a part of it.



All my best to each of you-

george

[geots]

The match has concluded for the Group 12 champion. Since I called it before it was over- which wasn't much of a risk- I thought I should stay up a few minutes longer and post the results. Let me just say that each final crosstable also shows the elo difference of the match. Maybe not a good idea, because 24 games will show many inflated elo differences. Just overlook that. Only thing that matters is the winner- not the margin he won by.



Inspiron 620 Intel i5 (SP8)-4 True Cores [40/21]
XPS 8500 Intel i5-4 True Cores [40/19]

Fritz 11 gui/Fritz 13 gui
1CPU/64-bit where available
128MB hash
Bases=NONE
Ponder_Learning=OFF
Perfect 2012b.ctg w/12-move limit
40/21 Repeating OR
40/19 Repeating (Benched to adapt to 40/40)
24-Game Matches

Code: Select all

Group 12 SP8-Core-i5  2013  

                                
Capivara LK 0.08b03 64-bit   +104    +12/-5/=7   64.58%   15.5/24
Shallow 1.0 64-bit           -104    +5/-12/=7   35.42%    8.5/24

Well, this is the "official" announcement:


Group 12- Capivara LK 0.08b03 64-bit


Congratulations to Capivara and his author. And I am ready now to test another version or 2 that has been released since these games started back in Sept.



Good nite all-

george

[Ajedrecista]

Hello Sven:

Code: Select all

Group 13 XPS-Core-i5     2013

KnockOut 0.7.1     +8/-4/=6   61.11%   11.0/18
Parrot 07.07.22    +4/-8/=6   38.89%    7.0/18

This one I will make an "unofficial call" on. I don't believe Parrot can make up 4 points in 6 games. KnockOut will be the Group 13 champion. Where are you, Sven?!

I am here

Honestly, I did not even expect that KnockOut would participate in any recent tournament since I did not release any updates during the past 2,5 years (reason being that I stopped its development and work on something new). So many thanks for including KnockOut in your tourney!

As to the match results, well, I think that beating any of the engines from the groups above 13 will be quite tough for KnockOut, therefore I would not expect any miracles.

EDIT: Of course the match against Parrot is not decided yet, although the probability for Parrot to get at least 5 points in 6 games is around 4% only, assuming 33% draws and equal strength.

Sven

Your post gave me an idea: write a Fortran 95 programme that calculates probabilities of finishing with certain scores; I used trinomial distributions and I struggled a lot with the programming (I am not a programmer) but finally I have something decent in less than 300 lines. I still have a minor problem: my programme refuses to do its job with some input data, such as draw ratios with some decimals... I do not know why. By the way, here is the complete trinomial distribution for KnockOut vs. Parrot:

Code: Select all

Probabilities_in_a_trinomial_distribution, ® 2013.

--------------------------------------------------------------------
Probabilities of all possible scores in a match between two engines.
--------------------------------------------------------------------

Write down the number of games of the match (up to 24):

6

Rating difference (between -800 and 800):

0

Probability of a draw (%) between 0 % and 100.00 %

33

Write down the clock rate of the CPU (in GHz), only for timing the elapsed time of the calculations:

3

End of the calculations. Approximated elapsed time:   15 ms.

Thanks for using Probabilities_in_a_trinomial_distribution. Press Enter to exit.

Here is the output notepad Probabilities.txt:

Code: Select all

Probabilities for a match of  6 games (rounded up to 0.0001%):
 
Rating difference (rounded up to 0.01 Elo):    0.00 Elo.
 
Probability of a win  = W ~  33.5000 %
Probability of a draw = D ~  33.0000 %
Probability of a lose = L ~  33.5000 %
 
----------------------------------------------------
 
Points:  6.0/ 6
 
+ 6 = 0 - 0
P ~   0.1413 %
 
Probability of win  6.0 points out of  6:   0.1413 %
 
----------------------------------------------------
 
Points:  5.5/ 6
 
+ 5 = 1 - 0
P ~   0.8354 %
 
Probability of win  5.5 points out of  6:   0.8354 %
 
----------------------------------------------------
 
Points:  5.0/ 6
 
+ 4 = 2 - 0
P ~   2.0573 %
 
+ 5 = 0 - 1
P ~   0.8480 %
 
Probability of win  5.0 points out of  6:   2.9054 %
 
----------------------------------------------------
 
Points:  4.5/ 6
 
+ 3 = 3 - 0
P ~   2.7021 %
 
+ 4 = 1 - 1
P ~   4.1769 %
 
Probability of win  4.5 points out of  6:   6.8791 %
 
----------------------------------------------------
 
Points:  4.0/ 6
 
+ 2 = 4 - 0
P ~   1.9963 %
 
+ 3 = 2 - 1
P ~   8.2292 %
 
+ 4 = 0 - 2
P ~   2.1201 %
 
Probability of win  4.0 points out of  6:  12.3457 %
 
----------------------------------------------------
 
Points:  3.5/ 6
 
+ 1 = 5 - 0
P ~   0.7866 %
 
+ 2 = 3 - 1
P ~   8.1064 %
 
+ 3 = 1 - 2
P ~   8.3539 %
 
Probability of win  3.5 points out of  6:  17.2469 %
 
----------------------------------------------------
 
Points:  3.0/ 6
 
+ 0 = 6 - 0
P ~   0.1291 %
 
+ 1 = 4 - 1
P ~   3.9927 %
 
+ 2 = 2 - 2
P ~  12.3438 %
 
+ 3 = 0 - 3
P ~   2.8268 %
 
Probability of win  3.0 points out of  6:  19.2925 %
 
----------------------------------------------------
 
Points:  2.5/ 6
 
+ 0 = 5 - 1
P ~   0.7866 %
 
+ 1 = 3 - 2
P ~   8.1064 %
 
+ 2 = 1 - 3
P ~   8.3539 %
 
Probability of win  2.5 points out of  6:  17.2469 %
 
----------------------------------------------------
 
Points:  2.0/ 6
 
+ 0 = 4 - 2
P ~   1.9963 %
 
+ 1 = 2 - 3
P ~   8.2292 %
 
+ 2 = 0 - 4
P ~   2.1201 %
 
Probability of win  2.0 points out of  6:  12.3457 %
 
----------------------------------------------------
 
Points:  1.5/ 6
 
+ 0 = 3 - 3
P ~   2.7021 %
 
+ 1 = 1 - 4
P ~   4.1769 %
 
Probability of win  1.5 points out of  6:   6.8791 %
 
----------------------------------------------------
 
Points:  1.0/ 6
 
+ 0 = 2 - 4
P ~   2.0573 %
 
+ 1 = 0 - 5
P ~   0.8480 %
 
Probability of win  1.0 points out of  6:   2.9054 %
 
----------------------------------------------------
 
Points:  0.5/ 6
 
+ 0 = 1 - 5
P ~   0.8354 %
 
Probability of win  0.5 points out of  6:   0.8354 %
 
----------------------------------------------------
 
Points:  0.0/ 6
 
+ 0 = 0 - 6
P ~   0.1413 %
 
Probability of win  0.0 points out of  6:   0.1413 %

Of course we agree in the numbers. It is a pity that my programme does not work always! The limit of 24 games is low but it seems that it does not work with more games (it worked with 25 but not with 26...). Well, I always write clumsy programmes... (sigh). At least my results are correct: I tried with a higher number of games (twelve IIRC) and what I called 'Probability of win X points out of Y' was almost normally distributed, so I guess that calculations are well done.

Regards from Spain.

Ajedrecista.

[geots]

Hello Sven:

Code: Select all

Group 13 XPS-Core-i5     2013

KnockOut 0.7.1     +8/-4/=6   61.11%   11.0/18
Parrot 07.07.22    +4/-8/=6   38.89%    7.0/18

This one I will make an "unofficial call" on. I don't believe Parrot can make up 4 points in 6 games. KnockOut will be the Group 13 champion. Where are you, Sven?!

I am here

Honestly, I did not even expect that KnockOut would participate in any recent tournament since I did not release any updates during the past 2,5 years (reason being that I stopped its development and work on something new). So many thanks for including KnockOut in your tourney!

As to the match results, well, I think that beating any of the engines from the groups above 13 will be quite tough for KnockOut, therefore I would not expect any miracles.

EDIT: Of course the match against Parrot is not decided yet, although the probability for Parrot to get at least 5 points in 6 games is around 4% only, assuming 33% draws and equal strength.

Sven

Your post gave me an idea: write a Fortran 95 programme that calculates probabilities of finishing with certain scores; I used trinomial distributions and I struggled a lot with the programming (I am not a programmer) but finally I have something decent in less than 300 lines. I still have a minor problem: my programme refuses to do its job with some input data, such as draw ratios with some decimals... I do not know why. By the way, here is the complete trinomial distribution for KnockOut vs. Parrot:

Code: Select all

Probabilities_in_a_trinomial_distribution, ® 2013.

--------------------------------------------------------------------
Probabilities of all possible scores in a match between two engines.
--------------------------------------------------------------------

Write down the number of games of the match (up to 24):

6

Rating difference (between -800 and 800):

0

Probability of a draw (%) between 0 % and 100.00 %

33

Write down the clock rate of the CPU (in GHz), only for timing the elapsed time of the calculations:

3

End of the calculations. Approximated elapsed time:   15 ms.

Thanks for using Probabilities_in_a_trinomial_distribution. Press Enter to exit.

Here is the output notepad Probabilities.txt:

Code: Select all

Probabilities for a match of  6 games (rounded up to 0.0001%):
 
Rating difference (rounded up to 0.01 Elo):    0.00 Elo.
 
Probability of a win  = W ~  33.5000 %
Probability of a draw = D ~  33.0000 %
Probability of a lose = L ~  33.5000 %
 
----------------------------------------------------
 
Points:  6.0/ 6
 
+ 6 = 0 - 0
P ~   0.1413 %
 
Probability of win  6.0 points out of  6:   0.1413 %
 
----------------------------------------------------
 
Points:  5.5/ 6
 
+ 5 = 1 - 0
P ~   0.8354 %
 
Probability of win  5.5 points out of  6:   0.8354 %
 
----------------------------------------------------
 
Points:  5.0/ 6
 
+ 4 = 2 - 0
P ~   2.0573 %
 
+ 5 = 0 - 1
P ~   0.8480 %
 
Probability of win  5.0 points out of  6:   2.9054 %
 
----------------------------------------------------
 
Points:  4.5/ 6
 
+ 3 = 3 - 0
P ~   2.7021 %
 
+ 4 = 1 - 1
P ~   4.1769 %
 
Probability of win  4.5 points out of  6:   6.8791 %
 
----------------------------------------------------
 
Points:  4.0/ 6
 
+ 2 = 4 - 0
P ~   1.9963 %
 
+ 3 = 2 - 1
P ~   8.2292 %
 
+ 4 = 0 - 2
P ~   2.1201 %
 
Probability of win  4.0 points out of  6:  12.3457 %
 
----------------------------------------------------
 
Points:  3.5/ 6
 
+ 1 = 5 - 0
P ~   0.7866 %
 
+ 2 = 3 - 1
P ~   8.1064 %
 
+ 3 = 1 - 2
P ~   8.3539 %
 
Probability of win  3.5 points out of  6:  17.2469 %
 
----------------------------------------------------
 
Points:  3.0/ 6
 
+ 0 = 6 - 0
P ~   0.1291 %
 
+ 1 = 4 - 1
P ~   3.9927 %
 
+ 2 = 2 - 2
P ~  12.3438 %
 
+ 3 = 0 - 3
P ~   2.8268 %
 
Probability of win  3.0 points out of  6:  19.2925 %
 
----------------------------------------------------
 
Points:  2.5/ 6
 
+ 0 = 5 - 1
P ~   0.7866 %
 
+ 1 = 3 - 2
P ~   8.1064 %
 
+ 2 = 1 - 3
P ~   8.3539 %
 
Probability of win  2.5 points out of  6:  17.2469 %
 
----------------------------------------------------
 
Points:  2.0/ 6
 
+ 0 = 4 - 2
P ~   1.9963 %
 
+ 1 = 2 - 3
P ~   8.2292 %
 
+ 2 = 0 - 4
P ~   2.1201 %
 
Probability of win  2.0 points out of  6:  12.3457 %
 
----------------------------------------------------
 
Points:  1.5/ 6
 
+ 0 = 3 - 3
P ~   2.7021 %
 
+ 1 = 1 - 4
P ~   4.1769 %
 
Probability of win  1.5 points out of  6:   6.8791 %
 
----------------------------------------------------
 
Points:  1.0/ 6
 
+ 0 = 2 - 4
P ~   2.0573 %
 
+ 1 = 0 - 5
P ~   0.8480 %
 
Probability of win  1.0 points out of  6:   2.9054 %
 
----------------------------------------------------
 
Points:  0.5/ 6
 
+ 0 = 1 - 5
P ~   0.8354 %
 
Probability of win  0.5 points out of  6:   0.8354 %
 
----------------------------------------------------
 
Points:  0.0/ 6
 
+ 0 = 0 - 6
P ~   0.1413 %
 
Probability of win  0.0 points out of  6:   0.1413 %

Of course we agree in the numbers. It is a pity that my programme does not work always! The limit of 24 games is low but it seems that it does not work with more games (it worked with 25 but not with 26...). Well, I always write clumsy programmes... (sigh). At least my results are correct: I tried with a higher number of games (twelve IIRC) and what I called 'Probability of win X points out of Y' was almost normally distributed, so I guess that calculations are well done.

Regards from Spain.

Ajedrecista.

Jesus, thanks for showing some interest. But one thing you said was "I always write clumsy programmes....." Just as sure as the sun will rise in the east, I wish to God I had the ability, brains and expertise to write "clumsy" programs such as yours.



You are a good friend- all the best,

george

[geots]

The following matches will begin within the hour:


Group 10
---------
Flux 2.2
Mephisto Gideon Pro


Group 9
--------
Djinn 0.960.2 x64
Gosu 0.16


Group 8
--------
Murka 2 x64
Typhoon 1.00-348


Group 7
--------
iCE 0.3
Delphil 2.9g



These should be 4 interesting match-ups.

george