Help from the 8 pieces endgame database ?

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

Help from the 8 pieces endgame database ?

Post by TAILLE »

Hi Ed,

Do you accept to help us analysing some positions which need the 8 pieces database ?

if yes what is the result of the following position ?

Image
Black to move

Damy plays 26-31 37x26 12-17 26-21 17x26 27-21 18x29 21-16 23x32 16x07 13-18 07-01 18-23 01-12 32-37 12-17 37-41 17-08 41-46 08-13 19x08 30x10

Image
black to move

and we reach a quite difficult 8 pieces endgame.

Thank you for your answer
Gérard
Ed Gilbert
Posts: 860
Joined: Sat Apr 28, 2007 14:53
Real name: Ed Gilbert
Location: Morristown, NJ USA
Contact:

Post by Ed Gilbert »

Hi Gerard,
Do you accept to help us analysing some positions which need the 8 pieces database ?
Sure, I welcome the chance to do something useful with it.

Kingsrow sees an immediate database draw at the start position and wants to play 11-16. It continues like this:

[FEN "B:W22,25,27,28,30,32,33,35,37,41:B11,12,13,14,18,19,23,24,26."]
11-16 2. 22-17 12x21 3. 41-36 23-29 4. 28-23 19x39 5. 30x10

However, 26-31 is a white win. But your 5x3 ending position is a database draw.

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

Post by TAILLE »

Hi Ed.,
Ed Gilbert wrote: [FEN "B:W22,25,27,28,30,32,33,35,37,41:B11,12,13,14,18,19,23,24,26."]
11-16 2. 22-17 12x21 3. 41-36 23-29 4. 28-23 19x39 5. 30x10
-- Ed
Thank you for your answer. It seems that there is something wrong in your position. There is no white man on square 41. The position I am analysing is a 9x9 position and I am wondering if black can win.
Damy eliminates quickly 11-16 due to 37-31 etc... but I had some difficulties with the promising gambit 26-31. I guess white can obtain the draw but I was not quite sure.
Ed Gilbert wrote:But your 5x3 ending position is a database draw.
Now I am almost convinced that 26-31 is not sufficient to win.

BTW I hope Kingsrow would play 26-31 rather than 11-16, in order to create difficulties for the weaker opponent!
Gérard
BertTuyt
Posts: 1592
Joined: Wed Sep 01, 2004 19:42

Post by BertTuyt »

Ed,

just a question out of curiosity.
With the 8p database, for which positions in general do you now see theoretical game-values within the normal search time-constraint ( 9m - 9m, 10m - 10m).

Bert
Ed Gilbert
Posts: 860
Joined: Sat Apr 28, 2007 14:53
Real name: Ed Gilbert
Location: Morristown, NJ USA
Contact:

Post by Ed Gilbert »

Hi Gerard,

Sorry about the setup error. That happens every once in a while, which is the reason I like to paste the fen setup along with the result. I'm glad that you checked it.

Your position is a draw. Kingsrow sees a database draw from the 9x9 start position and plays the same moves that you give, at least for the first 10 or so.

-- Ed
Ed Gilbert
Posts: 860
Joined: Sat Apr 28, 2007 14:53
Real name: Ed Gilbert
Location: Morristown, NJ USA
Contact:

Post by Ed Gilbert »

Hi Bert,
BertTuyt wrote: just a question out of curiosity.
With the 8p database, for which positions in general do you now see theoretical game-values within the normal search time-constraint ( 9m - 9m, 10m - 10m).
I don't think I can make any general statement about how many pieces can be present and see a db result. It all depends on how many plies it takes for the pv and other lines to convert into 8 pieces or less (or 5men vs. 4men, which I'm also using).

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

Post by TAILLE »

Hi Ed.,
Ed Gilbert wrote:Hi Gerard,

Sorry about the setup error. That happens every once in a while, which is the reason I like to paste the fen setup along with the result. I'm glad that you checked it.

Your position is a draw. Kingsrow sees a database draw from the 9x9 start position and plays the same moves that you give, at least for the first 10 or so.

-- Ed
Thank you Ed. Now I am pretty sure !
Gérard
TAILLE
Posts: 968
Joined: Thu Apr 26, 2007 18:51
Location: FRANCE

Post by TAILLE »

Hi Ed.,

Image
White to play

I am wondering if Kingsrow can help me to solve the above problem.

The main variant seems the following :
1.22-17 13-18 2.28-22 16-21 3.22x13 21x12 4.25-20 14x25 5.13-09 24-30 6.09-04 19-24 7.27-22 30-34 8.40-35 24-30 9.35x33 34-40 10.22-18 12-17 11.18x29 40-44

Image
White to play

and Damy is not able to solve this endgame.

Can Kingsrow solve this problem ?
Gérard
Ed Gilbert
Posts: 860
Joined: Sat Apr 28, 2007 14:53
Real name: Ed Gilbert
Location: Morristown, NJ USA
Contact:

Post by Ed Gilbert »

Hi Gerard,

You ending position is a white win (W:WK4,29,32,33,37,38:B17,25,26,44.). But since it takes about a 20-minute search to see that, I cannot give a certain answer for the earlier positions. I analyzed from the start position and kingsrow picks the same moves as damy when searching for a minute or so at each position. So it is likely that the whole line is a white win, but not a certainty.

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

Post by TAILLE »

Ed Gilbert wrote:Hi Gerard,

You ending position is a white win (W:WK4,29,32,33,37,38:B17,25,26,44.). But since it takes about a 20-minute search to see that, I cannot give a certain answer for the earlier positions. I analyzed from the start position and kingsrow picks the same moves as damy when searching for a minute or so at each position. So it is likely that the whole line is a white win, but not a certainty.

-- Ed
Thank you Ed. I will then try to solve the endgame by launching a calculation during all the night but I guess it will not sufficient if you need 20 minutes with the 8 pieces db.
Gérard
TAILLE
Posts: 968
Joined: Thu Apr 26, 2007 18:51
Location: FRANCE

Post by TAILLE »

Hi Ed,
Ed Gilbert wrote:Hi Gerard,

You ending position is a white win (W:WK4,29,32,33,37,38:B17,25,26,44.). But since it takes about a 20-minute search to see that, I cannot give a certain answer for the earlier positions. I analyzed from the start position and kingsrow picks the same moves as damy when searching for a minute or so at each position. So it is likely that the whole line is a white win, but not a certainty.

-- Ed
No progress for this endgame even after 8 hours calculation during the night.
Image
Damy began by 29-24 44-49 33-28 49-40 and did not find the win.
As a human I do not like the first move 29-24. By forcing Damy to chose 04-13 (instead of 29-24) Damy continues by 44-49 33-28 49-35 13-36 17-21 29-23 25-30 23-19 35-40 19-14 30-34 38-33 21-27 33x04 40-49 04-27 49-35 14-10 34-40 27-16 40-45 16-02 W+
Image
Could you please verify this sequence ?
Could you tell me if the first move 29-24 is wrong ? If not how is it possible to win after this move ?
Thank you in advance
Gérard
Ed Gilbert
Posts: 860
Joined: Sat Apr 28, 2007 14:53
Real name: Ed Gilbert
Location: Morristown, NJ USA
Contact:

Post by Ed Gilbert »

Hi Gerard,

4-13 is probably a draw.

[FEN "W:WK4,29,32,33,37,38:B17,25,26,44."]
1. 4-13 44-49 2. 33-28 49-35 3. 13-36 17-21 4. 29-23 35-44 {instead of damy's 25-30, which is a ww} 5. 36-22 44-35 6. 22-4 35-30 7. 4-36 30-39 8. 38-33 39-30 9. 28-22 30-8 {draw}

29-24 is the correct move to win. Maybe this means than unlike with Johan we don't have to worry about Gerard giving move advice to damy during a tournament :-)

[FEN "W:WK4,29,32,33,37,38:B17,25,26,44."]
1. 29-24 44-49 2. 24-19 (A) 49-40 3. 19-14 25-30 4. 14-10 30-35 5. 10-5 40-7 6. 5-28 {... ww}

(A) 2. 33-28 49-40 3. 4-36 40-45 4. 32-27 45-34 5. 37-32 34-1 6. 36-47 1-40 7. 24-19 40-49 8. 19-14 25-30 9. 14-10 30-35 {draw}

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

Post by TAILLE »

Ed Gilbert wrote:Hi Gerard,

4-13 is probably a draw.

[FEN "W:WK4,29,32,33,37,38:B17,25,26,44."]
1. 4-13 44-49 2. 33-28 49-35 3. 13-36 17-21 4. 29-23 35-44 {instead of damy's 25-30, which is a ww} 5. 36-22 44-35 6. 22-4 35-30 7. 4-36 30-39 8. 38-33 39-30 9. 28-22 30-8 {draw}

29-24 is the correct move to win. Maybe this means than unlike with Johan we don't have to worry about Gerard giving move advice to damy during a tournament

[FEN "W:WK4,29,32,33,37,38:B17,25,26,44."]
1. 29-24 44-49 2. 24-19 (A) 49-40 3. 19-14 25-30 4. 14-10 30-35 5. 10-5 40-7 6. 5-28 {... ww}

(A) 2. 33-28 49-40 3. 4-36 40-45 4. 32-27 45-34 5. 37-32 34-1 6. 36-47 1-40 7. 24-19 40-49 8. 19-14 25-30 9. 14-10 30-35 {draw}

-- Ed
Thank you very much Ed. for your help. Every body can clearly see the power of the 8 pieces db. Very good motivation for me to generate it in the future.
Just a last question : what would have played Kingsrow with only the 7 pieces endgame db ?
Gérard
Ed Gilbert
Posts: 860
Joined: Sat Apr 28, 2007 14:53
Real name: Ed Gilbert
Location: Morristown, NJ USA
Contact:

Post by Ed Gilbert »

Just a last question : what would have played Kingsrow with only the 7 pieces endgame db ?
Here are kingsrow's moves with the 7pc db using 60 second searches.

[FEN "W:WK4,29,32,33,37,38:B17,25,26,44."]
1. 29-24 44-49 2. 24-19 25-30 3. 19-14 30-35 4. 14-10 49-40 5. 4-9 40-45 6. 10-5

After the last move kingsrow sees a database win. From the search log, and using the 2.4GHz quad core with 6gb for cache buffers:

Code: Select all

best 35-40, value -200, depth 1/1.4/2, nodes 19, time 0.00, 19 kN/s, db 0
best 35-40, value -200, depth 3/4.7/10, nodes 290, time 0.00, 290 kN/s, db 31
best 17-21, value -280, depth 5/5.7/12, nodes 3915, time 0.00, 3915 kN/s, db 543
best 45-1, value -302, depth 7/7.0/15, nodes 27799, time 0.02, 1737 kN/s, db 2620
best 45-12, value -308, depth 9/9.4/18, nodes 133710, time 0.05, 2845 kN/s, db 15451
best 45-34, value -312, depth 11/11.4/21, nodes 641797, time 0.19, 3414 kN/s, db 82590
best 45-7, value -384, depth 13/12.9/24, nodes 1181691, time 0.39, 3022 kN/s, db 183293
best 45-12, value -470, depth 15/14.6/26, nodes 2360610, time 0.77, 3082 kN/s, db 380876
best 45-12, value -474, depth 17/14.6/27, nodes 2967975, time 0.97, 3063 kN/s, db 493606
best 17-21, value -476, depth 19/16.3/30, nodes 4537669, time 1.45, 3123 kN/s, db 817107
best 17-21, value -558, depth 21/19.4/32, nodes 11030895, time 4.02, 2747 kN/s, db 2294553
best 45-12, value -558, depth 23/18.3/32, nodes 17228709, time 6.38, 2703 kN/s, db 3402132
best 35-40, value -632, depth 25/22.1/38, nodes 24079681, time 9.11, 2643 kN/s, db 4590421
best 45-34, value -2566, depth 27/20.7/42, nodes 45835599, time 20.78, 2206 kN/s, db 8469221
best 45-34, value -2576, depth 29/19.3/40, nodes 74833193, time 38.09, 1964 kN/s, db 18593718
best 45-1, value -2582, depth 31/23.0/40, nodes 137222705, time 88.86, 1544 kN/s, db 45837112
best 45-1, value -2584, depth 33/24.8/41, nodes 183895428, time 126.97, 1448 kN/s, db 65848458
Rein Halbersma
Posts: 1722
Joined: Wed Apr 14, 2004 16:04
Contact:

Post by Rein Halbersma »

Ed, can you confirm that the following 4 positions are a white win? (i.e. the 3 variations on the shown diagram with the black king on squares 2,3 or 4 instead of square 1: are these also white wins? And can you confirm that with white king 37 on 48 and the black king on the single corner diagonal, that the position is a draw then? These type of positions are listed in a book on endgames with the results indicated.

Image
White to move, black king on 1, 2, 3 or 4: W+
With WK37 on 48 and black king on single diagonal: =
Post Reply