Unsolvable position

Discussion about development of draughts in the time of computer and Internet.
Sidiki
Posts: 313
Joined: Thu Jan 15, 2015 16:28
Real name: Coulibaly Sidiki

Re: Unsolvable position

Post by Sidiki » Sun Jul 31, 2022 18:09

Hi Joost,
look carefully, i activated 5 pieces endgame database to show that he use endgame and at the same time disabled it if you look again, you will see "Any" selected.

This engine, i will be honest the first time that i earned about it, i was no so convinced of his supposed strenght.

But after managed to get his Android version that for the moment run on my Android #Infinix Hot 10 play # with 4Go of Ram at 400 Kn/s, for the tests i done yesterday against Kingsrow, Scan and Mobydam.

Perspectives for Idames after only on 3 games by engines with 3 minutes / 90 moves:
0 win, 0 losse, 3 draws against Kingsrow 1.63
0 win, 0 losse, 3 draws against Scan 3. 1
3 win, 0 losse, 0 draws against Mobydam

Rein Halbersma
Posts: 1720
Joined: Wed Apr 14, 2004 16:04
Contact:

Re: Unsolvable position

Post by Rein Halbersma » Mon Aug 01, 2022 00:28

BertTuyt wrote:
Fri Jul 29, 2022 18:03
Sidiki, in all honesty, i do not remember this position either.
When did i post it?
Could be that my age is blocking my memory :D

Bert
FTFY! It was a post by Gérard Taille in 2012 viewtopic.php?f=53&t=2157&p=96637

Sidiki
Posts: 313
Joined: Thu Jan 15, 2015 16:28
Real name: Coulibaly Sidiki

Re: Unsolvable position

Post by Sidiki » Mon Aug 01, 2022 02:29

Rein Halbersma wrote:
Mon Aug 01, 2022 00:28
BertTuyt wrote:
Fri Jul 29, 2022 18:03
Sidiki, in all honesty, i do not remember this position either.
When did i post it?
Could be that my age is blocking my memory :D

Bert
FTFY! It was a post by Gérard Taille in 2012 viewtopic.php?f=53&t=2157&p=96637
Hi Rein,

Thanks for the post, i remember now it was Taille, no Bert, sorry for my bad memory Bert :D

Rein, it seem that you are the administrator of our great forum.

The author of Idames called Ngandeu Boris, tried a long time ago to join this forum but, it's what he told, that the sign up haven't be validated yet .

Can you if possible solve this problem it's seem that we will learn more together.

I will share the games played against Kingsrow, Scan and Mobydam today.

God bless you all.

Thanks, Sidiki

Joost Buijs
Posts: 460
Joined: Wed May 04, 2016 11:45
Real name: Joost Buijs

Re: Unsolvable position

Post by Joost Buijs » Mon Aug 01, 2022 08:24

Sidiki wrote:
Sun Jul 31, 2022 18:09
Hi Joost,
look carefully, i activated 5 pieces endgame database to show that he use endgame and at the same time disabled it if you look again, you will see "Any" selected.

This engine, i will be honest the first time that i earned about it, i was no so convinced of his supposed strenght.

But after managed to get his Android version that for the moment run on my Android #Infinix Hot 10 play # with 4Go of Ram at 400 Kn/s, for the tests i done yesterday against Kingsrow, Scan and Mobydam.

Perspectives for Idames after only on 3 games by engines with 3 minutes / 90 moves:
0 win, 0 losse, 3 draws against Kingsrow 1.63
0 win, 0 losse, 3 draws against Scan 3. 1
3 win, 0 losse, 0 draws against Mobydam
Hi Sidiki,

In my view "Any" means "what ever is available", and that is something different than "None".

Like Bert already said you have to play a match with "IDames" against other strong engines to determine how strong it is, only 9 games is not enough, if you really want to know it's strength (with a low error margin) you have to play at least 30.000 games (preferably with different opening lines).

Joost

Sidiki
Posts: 313
Joined: Thu Jan 15, 2015 16:28
Real name: Coulibaly Sidiki

Re: Unsolvable position

Post by Sidiki » Mon Aug 01, 2022 12:26

Hi Joost,

The author by "Any" mean "None".
The engine Effectively not used EGDB to solve this problem
My goal was to show that we have, the draughts community, a new comer.

Everybody know on this forum that I encourage news programs, news ideas, and most thing EXCHANGE.

I encouraged the NNUE integration on draughts and continuy doing it. In one word i support any thing that permit to this forum to grow up because, to be honest international draughts competition stopped with the release of Scan source code.
Source code that permit to many programs to get in strenght.
The proof it's that only Krzysztof and soon me will do the computer tournaments.

Also, to stop any misunderstanding, I'm not pretending that Idames it's the best engine in the world or saying that Kingsrow, Scan, Damage, Ares, Dragon and the others are weaks.

If someone has been hurted by my post, forgive me.

Friendly, Sidiki.

BertTuyt
Posts: 1573
Joined: Wed Sep 01, 2004 19:42

Re: Unsolvable position

Post by BertTuyt » Mon Aug 01, 2022 12:42

Rein,

apparently your memory is best.
It is indeed a position which was tested by Gerard.
When you go back in time , you see the same discussion (also Ed joined at that time).
If you want to win matches or solve problems you need to optimize and tune the search parameters in a different way.
Having said that, as the program Kingsrow of Ed is used often for analysis, he could add a switch between normal search and problem optimized search, think many people would like that.

I will do a test with pruning switched off in Damage, and see if this results in faster problem solving.

Bert

Sidiki
Posts: 313
Joined: Thu Jan 15, 2015 16:28
Real name: Coulibaly Sidiki

Re: Unsolvable position

Post by Sidiki » Mon Aug 01, 2022 13:17

BertTuyt wrote:
Mon Aug 01, 2022 12:42
Rein,

apparently your memory is best.
It is indeed a position which was tested by Gerard.
When you go back in time , you see the same discussion (also Ed joined at that time).
If you want to win matches or solve problems you need to optimize and tune the search parameters in a different way.
Having said that, as the program Kingsrow of Ed is used often for analysis, he could add a switch between normal search and problem optimized search, think many people would like that.

I will do a test with pruning switched off in Damage, and see if this results in faster problem solving.

Bert
Hi Bert,

To be honest, it's what i was waiting for.
Permit to everyone to solve this problem whith his engine without lose strenght.

Friendly, Sidiki

Ed Gilbert
Posts: 854
Joined: Sat Apr 28, 2007 14:53
Real name: Ed Gilbert
Location: Morristown, NJ USA
Contact:

Re: Unsolvable position

Post by Ed Gilbert » Mon Aug 01, 2022 13:30

Having said that, as the program Kingsrow of Ed is used often for analysis, he could add a switch between normal search and problem optimized search, think many people would like that.
The 8x8 versions of kingsrow have a control called "solve mode". When enabled, kingsrow is a lot better at solving (usually contrived) problems where you have to pitch a bunch of pieces first to gain an advantage. Solve mode turns off reductions and truncations that are based on beta cutoffs with a raised beta. It does not turn off LMR. In my experience LMR actually helps solve most of these kinds of problems.

However I don't think solve mode is useful for normal analysis, where you just want to analyze a game to find mistakes, etc. You need those reductions and truncations to get the most powerful analysis. It may not see some very deep combinations, but the ones that are so deep that they can't be found with reductions and pruning enabled occur with very low probability. Solve mode is only useful if you know in advance, or at least suspect, that the position has one of those very deep combinations. If you use it for all analysis, you're just using a weaker engine.

I had mixed feelings about adding this control to kingsrow. There will be some users that don't understand its purpose, and will just turn it on all the time thinking that it makes the engine stronger. I've seen this happen with another special purpose setting, "allscores" mode.

-- Ed

BertTuyt
Posts: 1573
Joined: Wed Sep 01, 2004 19:42

Re: Unsolvable position

Post by BertTuyt » Mon Aug 01, 2022 14:30

Sidiki,

when i switch off pruning, Damage finds the move 37-31 at ply 20 after 6.5 seconds with 1 core and at 16 ply 0.3 sec with 8 cores (lazy SMP).
In both cases the evaluation is +1 piece.

It is evident that for problem solving and normal game/match play search optimization/tuning is different.
Or in other words we did not find an algorithm which serves both at the same time.....

Bert

BertTuyt
Posts: 1573
Joined: Wed Sep 01, 2004 19:42

Re: Unsolvable position

Post by BertTuyt » Mon Aug 01, 2022 14:44

Sidiki,

and another thing :D .
This combination is relatively difficult for engines, as the 35-30 move from white does not yield a capture for black, but a quiet move.
And this quiet move is normally a trigger for pruning reductions.
But it remains an interesting position.

There is an option to make the pruning more "intelligent" for these types of positions, but an algorithm which improves problem solving without deterioration of match results is not yet in sight.

Bert

Sidiki
Posts: 313
Joined: Thu Jan 15, 2015 16:28
Real name: Coulibaly Sidiki

Re: Unsolvable position

Post by Sidiki » Mon Aug 01, 2022 15:32

BertTuyt wrote:
Mon Aug 01, 2022 14:44
Sidiki,

and another thing :D .
This combination is relatively difficult for engines, as the 35-30 move from white does not yield a capture for black, but a quiet move.
And this quiet move is normally a trigger for pruning reductions.
But it remains an interesting position.

There is an option to make the pruning more "intelligent" for these types of positions, but an algorithm which improves problem solving without deterioration of match results is not yet in sight.

Bert
:D
Yes Bert,

I see the idea,

But it seem that we are misunderstood. The author, it seem that I'm his ambassador, don't coded his engine specially to solve this kind of problem.

The engine it's strong also in playing. And this's only the Android version running at, after good look, 300Kn/s average.

You and the others know me, i don't tell that an engine it's strong if itsn't the case.
Once again itsn't to say that another one it's stronger than another else.

Friendly, Sidiki.

EricvanDusseldorp
Posts: 80
Joined: Wed Jan 03, 2018 16:53
Real name: Eric van Dusseldorp

Re: Unsolvable position

Post by EricvanDusseldorp » Wed Aug 03, 2022 07:47

Who is the composer of this position? I remember Ton Sijbrands, or not?

EricvanDusseldorp
Posts: 80
Joined: Wed Jan 03, 2018 16:53
Real name: Eric van Dusseldorp

Re: Unsolvable position

Post by EricvanDusseldorp » Wed Aug 03, 2022 09:56

Kingsrow with the 8 pieces endgame database found on the 28th ply (14th move) and more then one hour the winning opening move 37-31!

But the endgame was reached after 25 ply's (12 and a half move). I wonder that the winning move 37-31 was not found after 25 ply's.

ildjarn
Posts: 1537
Joined: Tue Aug 22, 2006 15:38
Real name: Joost de Heer

Re: Unsolvable position

Post by ildjarn » Wed Aug 03, 2022 14:39

EricvanDusseldorp wrote:
Wed Aug 03, 2022 07:47
Who is the composer of this position? I remember Ton Sijbrands, or not?
Yes, Ton Sijbrands, in his Volkskrant column.
Lasst die Maschinen verhungern, Ihr Narren...
Lasst sie verrecken!
Schlagt sie tot -- die Maschinen!

Sidiki
Posts: 313
Joined: Thu Jan 15, 2015 16:28
Real name: Coulibaly Sidiki

Re: Unsolvable position

Post by Sidiki » Wed Aug 03, 2022 17:45

EricvanDusseldorp wrote:
Wed Aug 03, 2022 09:56
Kingsrow with the 8 pieces endgame database found on the 28th ply (14th move) and more then one hour the winning opening move 37-31!

But the endgame was reached after 25 ply's (12 and a half move). I wonder that the winning move 37-31 was not found after 25 ply's.
Very nice and what happened without 8 endgame database ?

For information, Idames it's under migration to Java toward C++ and you will have news about it.

The author he's implementing dxp protocol to play thousands of games.

Also the "Any" = "None" has been fixed, i will post the new video with 0 EGDB and 1 minute +10 secondes time

Friendly, Sidiki.

Post Reply