Computer Olympiad 2017 Draughts

Discussion about development of draughts in the time of computer and Internet.
Krzysztof Grzelak
Posts: 782
Joined: Thu Jun 20, 2013 17:16
Real name: Krzysztof Grzelak

Re: Computer Olympiad 2017 Draughts

Post by Krzysztof Grzelak » Fri Jul 07, 2017 11:02

Fabien Letouzey wrote:No: it was Scan 3, and I didn't change the program during the tournament.

However I remember that GWD (and maybe Dragon Draughts) was modified between the two days. IMO it makes the tournament results harder to interpret; for example Scan played an improved GWD compared to others, but I might remember wrong.
Thanks for the reply Fabien. I have noticed that some programs always maintain their high level on the Olympics.

Rein Halbersma
Posts: 1635
Joined: Wed Apr 14, 2004 16:04
Contact:

Re: Computer Olympiad 2017 Draughts

Post by Rein Halbersma » Fri Jul 07, 2017 11:13

Image
Moby Dam - Scan

Scan played 35... 7-12? here and white created breathing room with 36. 23-18!, and although it looks promising for a while, there is no black win according to Kingsrow.

Another approach would be to avoid square <12> and let white advance, e.g. 35... 1-6 36. 31-27 17-21 37. 27-22 7-11 38. 23-18 8-12 39. 18x7 11x2 and it is hard to find an escape route for white. Food for Damy :)

TAILLE
Posts: 968
Joined: Thu Apr 26, 2007 18:51
Location: FRANCE

Re: Computer Olympiad 2017 Draughts

Post by TAILLE » Fri Jul 07, 2017 16:46

Rein Halbersma wrote:Image
Moby Dam - Scan

Scan played 35... 7-12? here and white created breathing room with 36. 23-18!, and although it looks promising for a while, there is no black win according to Kingsrow.

Another approach would be to avoid square <12> and let white advance, e.g. 35... 1-6 36. 31-27 17-21 37. 27-22 7-11 38. 23-18 8-12 39. 18x7 11x2 and it is hard to find an escape route for white. Food for Damy :)
Hi Rein,

It is a honour for me to see your reference to Damy to analyse this still very difficult position in a game Mobydam - Scan!!!

Yes Rein 35...7-12 seems not the best move for Damy.

Your proposal 35... 1-6 36. 31-27 17-21 37. 27-22 7-11 38. 23-18 8-12 39. 18x7 11x2 is very interesting but after the logical continuation 32-28 6-11 35-30 24x35 29-24 20x29 34x23 15-20 40-34 20-24 44-40 35x44 39x50 11-17 22x11 16x7 28-22
Image
Black to play

Damy is able to prove (after very long minutes!) the draw

That does not meen that the position you mentionned is a draw and I see another possibility. The idea is to keep a man on square 1, to wait for a white man coming on square 22 and to organise an attack against 22.

My proposal is the following
35...17-21 36.31-27 7-11 37.27-22 11-17 38.22x11 16x7 39.32-28 7-12 40.28-22
Image
Black to paly

and now black switch to (14-19) 23x14 (20x9) 29x20 (15x24) and white 22 is in a very poor position
Gérard

Rein Halbersma
Posts: 1635
Joined: Wed Apr 14, 2004 16:04
Contact:

Re: Computer Olympiad 2017 Draughts

Post by Rein Halbersma » Sun Jul 09, 2017 13:10

TAILLE wrote:
Rein Halbersma wrote:Image
Moby Dam - Scan

Scan played 35... 7-12? here and white created breathing room with 36. 23-18!, and although it looks promising for a while, there is no black win according to Kingsrow.

Another approach would be to avoid square <12> and let white advance, e.g. 35... 1-6 36. 31-27 17-21 37. 27-22 7-11 38. 23-18 8-12 39. 18x7 11x2 and it is hard to find an escape route for white. Food for Damy :)
Hi Rein,

It is a honour for me to see your reference to Damy to analyse this still very difficult position in a game Mobydam - Scan!!!
You are welcome. I actually meant that it was a type of position well suited to you for optimizing Damy, and you seem to enjoy such analyses. With Kingsrow + 8 piece db the quality of analysis is rather similar, but my time is limited.

TAILLE
Posts: 968
Joined: Thu Apr 26, 2007 18:51
Location: FRANCE

Re: Computer Olympiad 2017 Draughts

Post by TAILLE » Sun Jul 09, 2017 16:54

Hi,

Game Cerberus - Sjende Blyn
Image
Black to play

In the game black played 17-22 and did not manage to win.

Damy proposes 15-20 with the following PV
(15-20) 32-27 (15-20) 30-25 (7-11) 16x7 (12x1) 21x12 (18x7) 36-31 (33-38) 44-39 (7-12) with probably a win.
What is your view?
Gérard

Jelle Wiersma
Posts: 47
Joined: Wed Aug 12, 2015 17:45
Real name: Jelle Wiersma

Re: Computer Olympiad 2017 Draughts

Post by Jelle Wiersma » Sun Jul 09, 2017 19:13

TAILLE wrote: Game Cerberus - Sjende Blyn
In the game black played 17-22 and did not manage to win.
Damy proposes 15-20 with the following PV
(15-20) 32-27 (15-20) 30-25 (7-11) 16x7 (12x1) 21x12 (18x7) 36-31 (33-38) 44-39 (7-12) with probably a win.
I'm not sure it's a win, but after 62 seconds Sjende Blyn switches from 17-22 to 15-20 (value -1.3 man).
Just too much time for a move during these rapid games...

TAILLE
Posts: 968
Joined: Thu Apr 26, 2007 18:51
Location: FRANCE

Re: Computer Olympiad 2017 Draughts

Post by TAILLE » Sun Jul 09, 2017 20:05

Jelle Wiersma wrote:
TAILLE wrote: Game Cerberus - Sjende Blyn
In the game black played 17-22 and did not manage to win.
Damy proposes 15-20 with the following PV
(15-20) 32-27 (15-20) 30-25 (7-11) 16x7 (12x1) 21x12 (18x7) 36-31 (33-38) 44-39 (7-12) with probably a win.
I'm not sure it's a win, but after 62 seconds Sjende Blyn switches from 17-22 to 15-20 (value -1.3 man).
Just too much time for a move during these rapid games...
Hi Jelle,

Are you interested by knowing if (15-20) is a winning move or not?
If yes did you find another white move in the line above on which white might reach a draw?
Gérard

Jelle Wiersma
Posts: 47
Joined: Wed Aug 12, 2015 17:45
Real name: Jelle Wiersma

Re: Computer Olympiad 2017 Draughts

Post by Jelle Wiersma » Mon Jul 10, 2017 19:19

TAILLE wrote:
Jelle Wiersma wrote:
TAILLE wrote: Game Cerberus - Sjende Blyn
In the game black played 17-22 and did not manage to win.
Damy proposes 15-20 with the following PV
(15-20) 32-27 (15-20) 30-25 (7-11) 16x7 (12x1) 21x12 (18x7) 36-31 (33-38) 44-39 (7-12) with probably a win.
I'm not sure it's a win, but after 62 seconds Sjende Blyn switches from 17-22 to 15-20 (value -1.3 man).
Just too much time for a move during these rapid games...
Hi Jelle,

Are you interested by knowing if (15-20) is a winning move or not?
If yes did you find another white move in the line above on which white might reach a draw?
It's a bit different: ... 15-20; 32-27 7-11; 16x7 12x1; 21x12 18x7; 30-25 33-38; 44-39 23-28; 45-40 and indeed likely for black to win, but still not for sure

Rein Halbersma
Posts: 1635
Joined: Wed Apr 14, 2004 16:04
Contact:

Re: Computer Olympiad 2017 Draughts

Post by Rein Halbersma » Mon Jul 10, 2017 20:08

Hi Jelle,

Sjende Blyn is getting beter every year, very impressive! What's your secret? Can you tel us a bit more about the implementation? (Programming language, bitboard vs arrays, learning eval? which databases? etc., And how sid the program evolve?)

Rein

Jelle Wiersma
Posts: 47
Joined: Wed Aug 12, 2015 17:45
Real name: Jelle Wiersma

Re: Computer Olympiad 2017 Draughts

Post by Jelle Wiersma » Mon Jul 10, 2017 21:48

Rein Halbersma wrote:Hi Jelle,

Sjende Blyn is getting beter every year, very impressive! What's your secret? Can you tel us a bit more about the implementation? (Programming language, bitboard vs arrays, learning eval? which databases? etc., And how sid the program evolve?)

Rein
Hi Rein, that are a lot of questions!
That Sjende Blyn is getting better is because I'm working on it on a regular (and irregular!) basis.
It's written in C++. The positional evaluation is handwritten and is partly bitboard, partly array based. The database is 6p.

Do you develop draughts- or chess programs?

Rein Halbersma
Posts: 1635
Joined: Wed Apr 14, 2004 16:04
Contact:

Re: Computer Olympiad 2017 Draughts

Post by Rein Halbersma » Tue Jul 11, 2017 09:26

Jelle Wiersma wrote:
Rein Halbersma wrote:Hi Jelle,

Sjende Blyn is getting beter every year, very impressive! What's your secret? Can you tel us a bit more about the implementation? (Programming language, bitboard vs arrays, learning eval? which databases? etc., And how sid the program evolve?)

Rein
Hi Rein, that are a lot of questions!
That Sjende Blyn is getting better is because I'm working on it on a regular (and irregular!) basis.
It's written in C++. The positional evaluation is handwritten and is partly bitboard, partly array based. The database is 6p.

Do you develop draughts- or chess programs?
Thanks for the info. To your question, the answer is yes and no: I am writing a C++ template library that you can use to play draughts with any existing rules / board combination. Eventually, it wil have a complete engine class, most of the ingredients are there but in varying degrees of completion.

It's open source, Boost licensed, available GitHub:
https://github.com/rhalbersma/dctl

TAILLE
Posts: 968
Joined: Thu Apr 26, 2007 18:51
Location: FRANCE

Re: Computer Olympiad 2017 Draughts

Post by TAILLE » Tue Jul 11, 2017 12:07

Jelle Wiersma wrote:
TAILLE wrote:
Jelle Wiersma wrote:
I'm not sure it's a win, but after 62 seconds Sjende Blyn switches from 17-22 to 15-20 (value -1.3 man).
Just too much time for a move during these rapid games...
Hi Jelle,

Are you interested by knowing if (15-20) is a winning move or not?
If yes did you find another white move in the line above on which white might reach a draw?
It's a bit different: ... 15-20; 32-27 7-11; 16x7 12x1; 21x12 18x7; 30-25 33-38; 44-39 23-28; 45-40 and indeed likely for black to win, but still not for sure
No doubt for me that the line you show here above is surely a winning one: though Damy has not proved strictly the win I could see that the evaluation reached +4 after 30' and +7 after 8 hours (I was not able to wait more time because I got out of memory). Anyway no real risk for me to claim for a black win!
Gérard

TAILLE
Posts: 968
Joined: Thu Apr 26, 2007 18:51
Location: FRANCE

Re: Computer Olympiad 2017 Draughts

Post by TAILLE » Tue Jul 11, 2017 13:47

Hi Jelle,

The notation of the game Sjende Blyn - Deep Nikita ends on toernooibase in the following position

Image
Black to play

What was the endgame moves?
Gérard

Jelle Wiersma
Posts: 47
Joined: Wed Aug 12, 2015 17:45
Real name: Jelle Wiersma

Re: Computer Olympiad 2017 Draughts

Post by Jelle Wiersma » Tue Jul 11, 2017 20:03

TAILLE wrote:Hi Jelle,

The notation of the game Sjende Blyn - Deep Nikita ends on toernooibase in the following position

Image
Black to play

What was the endgame moves?
A few moves further in this game (and Deep Nikita in more trouble than it is here if I recall correctly) Andrew made a mistake in entering a move, which was discovered a little later when Sjende Blyn went for an "impossible" combination. He tried to correct it, but did not succeed. In the process, I reverted several moves played in the game, so all the moves played from this position were lost.
Andrew must have had a bad night; in the begin of the game he also managed to perform several moves played by Sjende Blyn (...) on the board and the clock... :?

Jelle Wiersma
Posts: 47
Joined: Wed Aug 12, 2015 17:45
Real name: Jelle Wiersma

Re: Computer Olympiad 2017 Draughts

Post by Jelle Wiersma » Tue Jul 11, 2017 20:06

TAILLE wrote:
Jelle Wiersma wrote:
TAILLE wrote:
Hi Jelle,

Are you interested by knowing if (15-20) is a winning move or not?
If yes did you find another white move in the line above on which white might reach a draw?
It's a bit different: ... 15-20; 32-27 7-11; 16x7 12x1; 21x12 18x7; 30-25 33-38; 44-39 23-28; 45-40 and indeed likely for black to win, but still not for sure
No doubt for me that the line you show here above is surely a winning one: though Damy has not proved strictly the win I could see that the evaluation reached +4 after 30' and +7 after 8 hours (I was not able to wait more time because I got out of memory). Anyway no real risk for me to claim for a black win!
Ok, I will add this alternative to toernooibase.

Post Reply