A difficult endgame in the game Coulibali - Ndjofang

Discussion about development of draughts in the time of computer and Internet.
TAILLE
Posts: 968
Joined: Thu Apr 26, 2007 18:51
Location: FRANCE

Re: A difficult endgame in the game Coulibali - Ndjofang

Post by TAILLE » Fri Oct 07, 2016 00:44

TAILLE wrote:
BertTuyt wrote:Gerard, thats amazing...
Do you use a 7P or 8P DB?

Bert
Hi Bert,
I use of course the 8p db. Anyway with my previous algorithm I was not able to resolve completly the positions given above so it was a very good news for me to see such inexpected results.
Hi Bert,

Image
Black to play

To be more complete I try the resolution of the Woldouby position with only the 7p db.
With the relevant egdb already in the cache Damy needs now 16 seconds to prove the draw.
Gérard

Joost Buijs
Posts: 317
Joined: Wed May 04, 2016 11:45
Real name: Joost Buijs

Re: A difficult endgame in the game Coulibali - Ndjofang

Post by Joost Buijs » Fri Oct 07, 2016 12:39

TAILLE wrote:
Image
White to play

My PV is the following
1.35-30 24x35 2.28-22 23-28 3.32x12 21x41 4.12-7 19-23 5.7-1 41-46 6.1x29 46-28 7.38-32 28x6 8.29-1 6-17 9.32-27 17-3 10.1-7 3-21
11.27-22 26-31 12.7-23 21-8 13.23-7 13-19 14.33-28 31-36 15.7-23 8-3 16.23x5 3-12 17.22-17 12x26 18.34-29 26-48 19.39-33 25-30 20.29-23 48-42 and the db is reached
This result looks amazing, I have one question about something I don't understand though.
The last capture seems to be at move 17 (34 ply) and you reach the database at move 20 (40 ply), shouldn't you reach the database immediately after the last capture, or do you use a partial 8 piece database with some positions excluded?

Joost

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

Re: A difficult endgame in the game Coulibali - Ndjofang

Post by TAILLE » Fri Oct 07, 2016 14:46

Joost Buijs wrote:
TAILLE wrote:
Image
White to play

My PV is the following
1.35-30 24x35 2.28-22 23-28 3.32x12 21x41 4.12-7 19-23 5.7-1 41-46 6.1x29 46-28 7.38-32 28x6 8.29-1 6-17 9.32-27 17-3 10.1-7 3-21
11.27-22 26-31 12.7-23 21-8 13.23-7 13-19 14.33-28 31-36 15.7-23 8-3 16.23x5 3-12 17.22-17 12x26 18.34-29 26-48 19.39-33 25-30 20.29-23 48-42 and the db is reached
This result looks amazing, I have one question about something I don't understand though.
The last capture seems to be at move 17 (34 ply) and you reach the database at move 20 (40 ply), shouldn't you reach the database immediately after the last capture, or do you use a partial 8 piece database with some positions excluded?

Joost
Hi Joost

My 8p db is complete (except the 7x1 db I didn't build).
At move 17 we have reached a 9 pieces position and Damy has do continue exploring the variants.
After the 20th move we reach the following position:

Image

It is still a 9 pieces position but a capture is unavoidable at the next move with a definite result in the 8p egdb.
Gérard

Joost Buijs
Posts: 317
Joined: Wed May 04, 2016 11:45
Real name: Joost Buijs

Re: A difficult endgame in the game Coulibali - Ndjofang

Post by Joost Buijs » Fri Oct 07, 2016 15:14

TAILLE wrote:
Hi Joost

My 8p db is complete (except the 7x1 db I didn't build).
At move 17 we have reached a 9 pieces position and Damy has do continue exploring the variants.
After the 20th move we reach the following position:

Image

It is still a 9 pieces position but a capture is unavoidable at the next move with a definite result in the 8p egdb.
Gerard, Thanks!

Even more impressive, Since I don't have a usable GUI atm. I only looked at the PV and assumed that there were 8 pieces left after the last capture.
I don't think that my very immature program can reach these depths in a reasonable amount time, on that particular position it does 34 ply within a minute or so, but to reach 40 ply wo quiescence I need additional pruning or selective extensions.

Joost

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

Re: A difficult endgame in the game Coulibali - Ndjofang

Post by TAILLE » Fri Oct 07, 2016 15:52

Joost Buijs wrote:
TAILLE wrote:
Hi Joost

My 8p db is complete (except the 7x1 db I didn't build).
At move 17 we have reached a 9 pieces position and Damy has do continue exploring the variants.
After the 20th move we reach the following position:

Image

It is still a 9 pieces position but a capture is unavoidable at the next move with a definite result in the 8p egdb.
Gerard, Thanks!

Even more impressive, Since I don't have a usable GUI atm. I only looked at the PV and assumed that there were 8 pieces left after the last capture.
I don't think that my very immature program can reach these depths in a reasonable amount time, on that particular position it does 34 ply within a minute or so, but to reach 40 ply wo quiescence I need additional pruning or selective extensions.

Joost
Hi Joost,

As a comment, some moves in my PV were quite difficult to find. For example let's take the position after 4.12-7:

Image
Black to play

4...19-23!! is the only winning move. Is your program able to prove the win from this diagram or is depth still to large?
Gérard

Krzysztof Grzelak
Posts: 782
Joined: Thu Jun 20, 2013 17:16
Real name: Krzysztof Grzelak

Re: A difficult endgame in the game Coulibali - Ndjofang

Post by Krzysztof Grzelak » Fri Oct 07, 2016 16:57

I wonder how far the program has to analyze the position to demonstrate victory. Movement 19-23 give an advantage but is a winning move. You can on this subject a lot to discuss.
Last edited by Krzysztof Grzelak on Fri Oct 07, 2016 17:04, edited 1 time in total.

Joost Buijs
Posts: 317
Joined: Wed May 04, 2016 11:45
Real name: Joost Buijs

Re: A difficult endgame in the game Coulibali - Ndjofang

Post by Joost Buijs » Fri Oct 07, 2016 17:03

TAILLE wrote: Hi Joost,

As a comment, some moves in my PV were quite difficult to find. For example let's take the position after 4.12-7:

Image
Black to play

4...19-23!! is the only winning move. Is your program able to prove the win from this diagram or is depth still to large?
Hi Gerard,

My program doesn't use egdb's yet, a month ago I was busy building an egdb generator which is for a large part finished, after I started working on the search the work on the egdb came more or less to a standstill. However, I can temporary use Ed's egtb, I recently made a dll for it but it still has to be included in my code, this weekend I will try to add it.

The solution for the above position (when using 8p egdb) is 33 ply, with Ed's egtb added my program will probably find it within a few minutes, since I have no idea what the egdb will do to the tree-size it is all speculation, I will post the results later on.

Joost

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

Re: A difficult endgame in the game Coulibali - Ndjofang

Post by TAILLE » Fri Oct 07, 2016 23:23

Hi,

Because this discussion has now nothing to do with the initial subject I propose to continue on the subject "search algorithm" because we are really talking about a new search algorithm.
Gérard

Post Reply