Unofficial World Championship Of Computer Programs In International Draughts 2021

[Krzysztof Grzelak]

Unfortunately, here this page not work. Can it be realized on YouTube like last year?

I'm sorry, but there's nothing I can do about this now.

[Krzysztof Grzelak]

The tournament has already started very interesting...
In advance, congratulations Grzelak!

A question, are you using any opening books for Flits 3.1?
I realize that Flits would hardly lose some of the perts I've seen if he was without a book, for example.

I happen to have the version here, and I noticed that he doesn't usually play these moves. His book when activated is pretty incomplete, I ended up adapting the 1.5 version book to see if it gets stronger.

Book from Turbodambase 10.

[Joost Buijs]

After fixing the time control bug in my engine (this was a real bug) I noticed that the engine used more time than expected and was at risk losing on time. According to the tournament rules it is allowed to update the engine twice. So I asked Krzysztof to replace it with an updated version that has this problem fixed.

It can be downloaded here: https://www.dropbox.com/s/uf67rgluevb1e ... b.zip?dl=0

Joost

[pontel]

The tournament has already started very interesting...
In advance, congratulations Grzelak!

A question, are you using any opening books for Flits 3.1?
I realize that Flits would hardly lose some of the perts I've seen if he was without a book, for example.

I happen to have the version here, and I noticed that he doesn't usually play these moves. His book when activated is pretty incomplete, I ended up adapting the 1.5 version book to see if it gets stronger.

Book from Turbodambase 10.

It is a book based on Turbodambase's extensive game bank, but a human game book. It will always respond with the most played bid according to match history.

https://toernooibase.kndb.nl/applet/ope ... e2=0&taal=

If you play: 01. 32-28 the move 19-23 will always be the answer from Flits3.1/Turbodambase10. The same in turn will always play as white 01. 32-28.

[Joost Buijs]

After watching the tournament for some time I noticed that my engine (after the previous fix) now plays too fast.

According to tournament rules I could submit one more update, so I asked Krzysztof to replace it with version 1.43c which tries to improve this situation somewhat. It can be downloaded here: https://www.dropbox.com/s/jltsl0m22wial ... c.zip?dl=0

These things happened because I added a book and pondering just a few days before the tournament started and this disrupted the time-control more than expected.

Joost

[Jelle Wiersma]

You have long days Krzysztof! Appreciate it!
For those interested: Hendrik Veenstra has added this tournament and games to https://toernooibase.kndb.nl

[Krzysztof Grzelak]

There is a one-day break in the tournament tomorrow.

[pontel]

There is a one-day break in the tournament tomorrow.

Bom descanso meu amigo!

[Krzysztof Grzelak]

Tournament table of the tournament " The Unofficial World Championship Of Computer Programs In International Draughts 2021 " after 20 rounds. For more information, please visit the website https://www.uwccpi2021.draughtsprograms.eu/ The tournament continues tomorrow from 8:00 CEST.

[Joost Buijs]

Krzysztof, thank you for organizing this event!

The amount of effort needed to play all these games on a DGT E-board must be tremendous.

Despite their lower speed both NNUE programs seem to perform reasonably well.

Both programs use 'supervised learning' meaning their network learns from the outcome of games that have been played in advance. Although many games are used these games still contain errors which has an adversary effect on network quality.

Starting with a completely random network and letting it learn from scratch while playing 'reinforcement learning' will probably yield even better results. Certainly something that has to be tried in the near future.

[jj]

[Event "UWC 2021"]
[Date "2021.09.13"]
[Round "23"]
[White "Dragon"]
[Black "Ares"]
[Result "*"]
1. 32-28 18-22 2. 37-32 12-18 3. 32-27 19-23 4. 28x19 14x23 5. 41-37 7-12 6. 34-29 23x34 7. 40x29 13-19 8. 37-32 17-21 9. 31-26 22x31 10. 26x37 8-13 11. 36-31 21-26 12. 45-40 2-8 13. 39-34 10-14 14. 43-39 5-10 15. 46-41 11-17 16. 31-27 19-24 17. 27-21 16x27 18. 32x21 1-7 19. 37-32 17-22 20. 21-16 22-27 21. 32x21 26x17 22. 42-37 18-22 23. 37-32 22-27 24. 32x21 17x26 25. 41-37 6-11 26. 49-43 12-17 27. 38-32 8-12 28. 32-28 13-18 29. 43-38 3-8 30. 38-32 9-13 31. 32-27 17-21 32. 37-32 4-9 33. 50-45 14-19 34. 47-42 20-25 35. 29x20 25x14 36. 42-37 15-20 37. 34-30 18-23 38. 30-25 20-24 39. 48-42 10-15 40. 40-34 15-20 41. 34-30 12-18 42. 39-34 11-17 43. 34-29 23x34 44. 30x39 17-22 45. 28x17 21x12 46. 45-40 12-17 47. 32-28 18-22 48. 27x18 13x22 49. 40-34 9-13 50. 34-29 7-11 51. 16x7 17-21 52. 28x17 21x1 53. 42-38 8-12 54. 38-32 1-7 55. 39-34 13-18 56. 34-30 12-17 57. 44-39 17-22 58. 32-28 22-27 59. 28-23 19x28 60. 30x10 20-24 61. 33x13 24x44 62. 10-5 44-49 63. 37-31 27x36 64. 13-9 49-38 *


white to move

This interesting game, which I happened to see live today, was ended here and scored a draw. So I copied the moves to check. The position is clearly a win for White. Firstly, it is an 8 piece database win (Kingsrow) and secondly, you don't need an 8 piece database to win it. White easily gets 4 kings (Black can do nothing) and calculates the remaining win. Using a 7 or 6 piece database suffices.

The remaining games of round 23 were played but after that a new game Dragon-Ares was played, which ended in a draw. This game is now on the live site.

What happened?

[Krzysztof Grzelak]

[Event "UWC 2021"]
[Date "2021.09.13"]
[Round "23"]
[White "Dragon"]
[Black "Ares"]
[Result "*"]
1. 32-28 18-22 2. 37-32 12-18 3. 32-27 19-23 4. 28x19 14x23 5. 41-37 7-12 6. 34-29 23x34 7. 40x29 13-19 8. 37-32 17-21 9. 31-26 22x31 10. 26x37 8-13 11. 36-31 21-26 12. 45-40 2-8 13. 39-34 10-14 14. 43-39 5-10 15. 46-41 11-17 16. 31-27 19-24 17. 27-21 16x27 18. 32x21 1-7 19. 37-32 17-22 20. 21-16 22-27 21. 32x21 26x17 22. 42-37 18-22 23. 37-32 22-27 24. 32x21 17x26 25. 41-37 6-11 26. 49-43 12-17 27. 38-32 8-12 28. 32-28 13-18 29. 43-38 3-8 30. 38-32 9-13 31. 32-27 17-21 32. 37-32 4-9 33. 50-45 14-19 34. 47-42 20-25 35. 29x20 25x14 36. 42-37 15-20 37. 34-30 18-23 38. 30-25 20-24 39. 48-42 10-15 40. 40-34 15-20 41. 34-30 12-18 42. 39-34 11-17 43. 34-29 23x34 44. 30x39 17-22 45. 28x17 21x12 46. 45-40 12-17 47. 32-28 18-22 48. 27x18 13x22 49. 40-34 9-13 50. 34-29 7-11 51. 16x7 17-21 52. 28x17 21x1 53. 42-38 8-12 54. 38-32 1-7 55. 39-34 13-18 56. 34-30 12-17 57. 44-39 17-22 58. 32-28 22-27 59. 28-23 19x28 60. 30x10 20-24 61. 33x13 24x44 62. 10-5 44-49 63. 37-31 27x36 64. 13-9 49-38 *


white to move

This interesting game, which I happened to see live today, was ended here and scored a draw. So I copied the moves to check. The position is clearly a win for White. Firstly, it is an 8 piece database win (Kingsrow) and secondly, you don't need an 8 piece database to win it. White easily gets 4 kings (Black can do nothing) and calculates the remaining win. Using a 7 or 6 piece database suffices.

The remaining games of round 23 were played but after that a new game Dragon-Ares was played, which ended in a draw. This game is now on the live site.

What happened?

Yes that's right, I repeated the game for the sake of fairness. After the last move, Dragon stopped working. It was difficult for me to assess the position on the board right away. The regulations also do not say that when the drafts engine stops working, it loses the game. I think that I will make such a provision in the regulations in the future. I didn't want to hurt anyone here with the result of the game. There is still a problem with the endgame bases in Dragon. Michael (author of Dragon) does not want to share the 8 figure base.

[pontel]

[Event "UWC 2021"]
[Date "2021.09.13"]
[Round "23"]
[White "Dragon"]
[Black "Ares"]
[Result "*"]
1. 32-28 18-22 2. 37-32 12-18 3. 32-27 19-23 4. 28x19 14x23 5. 41-37 7-12 6. 34-29 23x34 7. 40x29 13-19 8. 37-32 17-21 9. 31-26 22x31 10. 26x37 8-13 11. 36-31 21-26 12. 45-40 2-8 13. 39-34 10-14 14. 43-39 5-10 15. 46-41 11-17 16. 31-27 19-24 17. 27-21 16x27 18. 32x21 1-7 19. 37-32 17-22 20. 21-16 22-27 21. 32x21 26x17 22. 42-37 18-22 23. 37-32 22-27 24. 32x21 17x26 25. 41-37 6-11 26. 49-43 12-17 27. 38-32 8-12 28. 32-28 13-18 29. 43-38 3-8 30. 38-32 9-13 31. 32-27 17-21 32. 37-32 4-9 33. 50-45 14-19 34. 47-42 20-25 35. 29x20 25x14 36. 42-37 15-20 37. 34-30 18-23 38. 30-25 20-24 39. 48-42 10-15 40. 40-34 15-20 41. 34-30 12-18 42. 39-34 11-17 43. 34-29 23x34 44. 30x39 17-22 45. 28x17 21x12 46. 45-40 12-17 47. 32-28 18-22 48. 27x18 13x22 49. 40-34 9-13 50. 34-29 7-11 51. 16x7 17-21 52. 28x17 21x1 53. 42-38 8-12 54. 38-32 1-7 55. 39-34 13-18 56. 34-30 12-17 57. 44-39 17-22 58. 32-28 22-27 59. 28-23 19x28 60. 30x10 20-24 61. 33x13 24x44 62. 10-5 44-49 63. 37-31 27x36 64. 13-9 49-38 *


white to move

This interesting game, which I happened to see live today, was ended here and scored a draw. So I copied the moves to check. The position is clearly a win for White. Firstly, it is an 8 piece database win (Kingsrow) and secondly, you don't need an 8 piece database to win it. White easily gets 4 kings (Black can do nothing) and calculates the remaining win. Using a 7 or 6 piece database suffices.

The remaining games of round 23 were played but after that a new game Dragon-Ares was played, which ended in a draw. This game is now on the live site.

What happened?

Yes that's right, I repeated the game for the sake of fairness. After the last move, Dragon stopped working. It was difficult for me to assess the position on the board right away. The regulations also do not say that when the drafts engine stops working, it loses the game. I think that I will make such a provision in the regulations in the future. I didn't want to hurt anyone here with the result of the game. There is still a problem with the endgame bases in Dragon. Michael (author of Dragon) does not want to share the 8 figure base.

In fact Dragon Drafts bases are a big problem. The simple fact of storing them already gave me two hard drives, and the worst thing is that even I have generated the bases of 7 complete pieces they contain many errors and they usually practice perpetual moves without developing the step by step for the victories. The program is usually more efficient with bases smaller than 4 pieces.

What maybe could be done, would be to Generate a PDN with the FEN in position so that the programs could follow the game with the reduced time.

[Ed Gilbert]

What maybe could be done, would be to Generate a PDN with the FEN in position so that the programs could follow the game with the reduced time.

Highlight the list of moves in the window next to the board display, and copy/paste them into kingsrow. Click anywhere in the kingsrow window to make it the foreground app and then CTRL-V. For example, like this list below which I just copied from a recent game.

01. 32-28 18-22
02. 37-32 12-18
03. 41-37 07-12
04. 46-41 22-27
05. 31x22 18x27
06. 32x21 17x26
07. 37-32 01-07
08. 41-37 12-18

edit: I should add, the list above is a valid pdn game. It doesn't have a * game terminator, but any pdn parser should handle it properly. Headers are not necessary.

[pontel]

What maybe could be done, would be to Generate a PDN with the FEN in position so that the programs could follow the game with the reduced time.

edit: I should add, the list above is a valid pdn game. It doesn't have a * game terminator, but any pdn parser should handle it properly. Headers are not necessary.

Hi Ed!

I'm not fluent in English and maybe the translator sabotaged me. What I tried to do was suggest to Krzysztof Grzelak to save the PDN with the FEN from position to continue the game where it left off.

Even so, thank you very much for your availability!