SF6 moves comparison: Event Vugar Gashimov Mem 2015

[Ferdy]

Quick comparison with Linares 1994.

I think Lineras had longer time controls, too, what with adjournments (which I think Lineras had). Modern players are fast as well as good.

It is also interesting to compare the older tournaments, but this really needs computer time. More depths is better of course. I am starting to collect some pgn files of these older popular tournaments.

If you ask, I can use my computer (6 cores @ 4 GHz) to analyze some old tournaments overnight. I'd prefer 1 min by move on 1 core (* 6).
I'd prefer tournament from '80s and '90. May be start with world championships and after that, top tournaments (Linares, Wijk-aan-zee, ...).

Right I will look for some interesting championship matches.

[Vinvin]

Quick comparison with Linares 1994.

I think Lineras had longer time controls, too, what with adjournments (which I think Lineras had). Modern players are fast as well as good.

It is also interesting to compare the older tournaments, but this really needs computer time. More depths is better of course. I am starting to collect some pgn files of these older popular tournaments.

If you ask, I can use my computer (6 cores @ 4 GHz) to analyze some old tournaments overnight. I'd prefer 1 min by move on 1 core (* 6).
I'd prefer tournament from '80s and '90. May be start with world championships and after that, top tournaments (Linares, Wijk-aan-zee, ...).

Right I will look for some interesting championship matches.

The match Adams (0.5) vs Hydra (5.5) should be interesting to evaluate too !
http://en.chessbase.com/post/adams-vs-h ... achine-5-5
What was the time control ? I don't remember if there was increments or not ...

[Ferdy]

Quick comparison with Linares 1994.

I think Lineras had longer time controls, too, what with adjournments (which I think Lineras had). Modern players are fast as well as good.

It is also interesting to compare the older tournaments, but this really needs computer time. More depths is better of course. I am starting to collect some pgn files of these older popular tournaments.

If you ask, I can use my computer (6 cores @ 4 GHz) to analyze some old tournaments overnight. I'd prefer 1 min by move on 1 core (* 6).
I'd prefer tournament from '80s and '90. May be start with world championships and after that, top tournaments (Linares, Wijk-aan-zee, ...).

Right I will look for some interesting championship matches.

The match Adams (0.5) vs Hydra (5.5) should be interesting to evaluate too !
http://en.chessbase.com/post/adams-vs-h ... achine-5-5
What was the time control ? I don't remember if there was increments or not ...

I am not sure too, it only says regular time control, at least time is longer.

I got the epd of this match now, there was some minor issues I fixed for pgn from matches, I did not get the correct win/loss/draw, but it has no effect on analysis of errors.

I am trying to improve the database saving of info from analysis, so analysis effort will not go to waste. Still the basic idea is separate each analysis sessions by tournaments, or matches. After which the generated database can then be combined. In the combined database we will be able to get specific info thru SQL statements. Here is a sample output from combined database from Adams-Hydra and Gashimov with analysis time of 10ms/pos just a quick test comparing the event average error, and the error from Hydra. Stockfish 6 is used.

Code: Select all

Players                     Event                         AveError  GMEveAveErr  AHEveAveErr
--------------------------  ----------------------------  --------  -----------  -----------
HYDRA                       Adams-Hydra,London2005        6.644329  14.47595558  13.18342149
Carlsen, Magnus             GashimovMemorial,Shamkir2015  9.314285  14.47595558  13.18342149
Giri, Anish                 GashimovMemorial,Shamkir2015  12.05405  14.47595558  13.18342149
So, Wesley                  GashimovMemorial,Shamkir2015  12.11469  14.47595558  13.18342149
Anand, Viswanathan          GashimovMemorial,Shamkir2015  12.16423  14.47595558  13.18342149
Mamedov, Rauf               GashimovMemorial,Shamkir2015  12.50196  14.47595558  13.18342149
Caruana, Fabiano            GashimovMemorial,Shamkir2015  13.09034  14.47595558  13.18342149
Mamedyarov, Shakhriyar      GashimovMemorial,Shamkir2015  15.06333  14.47595558  13.18342149
Vachier-Lagrave, Maxime     GashimovMemorial,Shamkir2015  16.72697  14.47595558  13.18342149
Adams, Michael              Adams-Hydra,London2005        19.72251  14.47595558  13.18342149
Adams, Michael              GashimovMemorial,Shamkir2015  20.068    14.47595558  13.18342149
Kramnik, Vladimir           GashimovMemorial,Shamkir2015  21.66167  14.47595558  13.18342149 

That match is of high quality and Adams is just doing fine. This is based on sf at 10ms of course.

The complete table.

Code: Select all

Players                     Event                         TotalPos  AveDepth  AveError  DECnt   E300CP  E200CP  E100CP  E50CP   E20CP   EECnt   W    L    D    Score  SRate   
--------------------------  ----------------------------  --------  --------  --------  ------  ------  ------  ------  ------  ------  ------  ---  ---  ---  -----  --------
HYDRA                       Adams-Hydra,London2005        194       11        6.644329  19      0       0       0       1       14      18      5    0    1     5.5   91.67%  
Carlsen, Magnus             GashimovMemorial,Shamkir2015  280       12        9.314285  31      0       0       0       5       24      23      5    0    4     7.0   77.78%  
Giri, Anish                 GashimovMemorial,Shamkir2015  296       12        12.05405  41      0       0       1       9       24      27      0    2    7     3.5   38.89%  
So, Wesley                  GashimovMemorial,Shamkir2015  279       11        12.11469  33      0       0       1       7       27      23      3    2    4     5.0   55.56%  
Anand, Viswanathan          GashimovMemorial,Shamkir2015  274       12        12.16423  38      0       0       1       13      26      28      3    0    6     6.0   66.67%  
Mamedov, Rauf               GashimovMemorial,Shamkir2015  255       12        12.50196  40      0       0       2       5       17      25      0    2    7     3.5   38.89%  
Caruana, Fabiano            GashimovMemorial,Shamkir2015  321       12        13.09034  48      0       0       1       9       22      23      2    1    6     5.0   55.56%  
Mamedyarov, Shakhriyar      GashimovMemorial,Shamkir2015  300       11        15.06333  42      1       0       1       7       20      28      1    2    6     4.0   44.44%  
Vachier-Lagrave, Maxime     GashimovMemorial,Shamkir2015  304       12        16.72697  60      0       2       1       3       22      25      0    2    7     3.5   38.89%  
Adams, Michael              Adams-Hydra,London2005        191       11        19.72251  56      0       1       0       4       15      11      0    5    1     0.5   8.33%   
Adams, Michael              GashimovMemorial,Shamkir2015  250       12        20.068    56      1       1       0       7       24      27      1    3    5     3.5   38.89%  
Kramnik, Vladimir           GashimovMemorial,Shamkir2015  334       12        21.66167  65      0       0       0       11      31      30      2    3    4     4.0   44.44%

There are missing info in this table that I am trying to address, save the hardware name where the analysis came from, save the analysis engine name and version, it could happen that different people will use different engine - looking ahead in the future. Save the analysis time, save the hash, threads. Also get info on the material of the positions being analyzed, perhaps there are players whose games will always go to the endings on average. Average position complexity can also be saved. Just get how many times the engine is changing its best move from iteration to iterations during analysis. Later querying all these things would be fun. Every analysis can be saved, short time or longer time does not matter. We can also saved all the positions that are analyzed, complete with depth, bestmove, analysis time, nodes etc, etc.

In that table above I use engine error margin of 10cp, meaning, if player score is better than 10cp, that will count as engine error in EECnt.

I am not going to release it yet, I will try to work on the database. Suggest what would be entered in the database, this is important as I am not going to come back into this after finishing the design.