World Draughts Forum
De Groen van Prinstererschool in Oud-Alblas doet mee met 20 DamZ! DamSets.
De Damsets worden door Mari Dame opgehaald bij DamZ! distributeur Jeroen Schakel in Brandwijk.
Statistics: Posted by Marcel Kosters — Tue Nov 24, 2020 22:21
What to do to make the engine continue to play after exceeding the moves for the game and there is no message about the end of the game. For example, I set the program to 15 minutes for 75 moves and after 75 moves it doesn't play anymore.
Statistics: Posted by Krzysztof Grzelak — Tue Nov 24, 2020 13:56
In Eindhoven doet Fred Passchier mee met 20 DamZ! DamSets voor Sinterklaas. Fred haalt de sets op bij DamZ! distributeur Piet van Erp in Sint Oedenrode.
Statistics: Posted by Marcel Kosters — Tue Nov 24, 2020 12:41
Leuk en goed idee over Jannes, Harm, Ton, Roel, Jan, Olga of een fictieve dam-ster: doen . Zelf heb ik voorlopig de handen vol aan DamZ!Afgelopen week bij M een aantal schakers op bezoek, omdat schaken weer helemaal hip is dankzij Netflix (The queen's gambit). Wie gaat Netflix overtuigen een serie over Jannes te maken?
Statistics: Posted by Marcel Kosters — Tue Nov 24, 2020 09:44
Statistics: Posted by ildjarn — Tue Nov 24, 2020 07:32
Statistics: Posted by Marcel Kosters — Mon Nov 23, 2020 14:56
Statistics: Posted by perepelkins — Mon Nov 23, 2020 13:25
Stsiapan Darashenka alone in the lead
After the fifth round of the Toernooibase Lidraughts Hybrid League Cup the Belarusian player Stsiapan Darashenka has the sole lead with 9 out of 5. He won yesterday against top seed (on rating) Erik van de Weerdhof. Van de Weerdhof walked into a trick and resigned after 19 moves.
Gerrit Draaisma played a draw against Wiel Maertzdorf and got one point behind. And Heorhi Vydzerka also got 8 out of 5, with a win against Julien Clément.
Six players have 7 points, with two players form Belarus and two players from Dama Latina. Wiel Maertzdorf and organizer Piet Bouma (lucky draw against Yauhen Zhukouski) are also in this group.
The team Belarus has now 31 points in het team classification. On quite a distance Various Netherlands (24), BelFraCam (24) and Dama Latina (23) fight for the second place. Frisian Flag (21) and Schollekoppen (19) have lost connection, and SDFOI closes the line-up with 13 points. One game between Mukul Thakur and Hanny de Vaal-Kokshoorn is not played yesterday. Maybe today they can complete round 5.
So in the individual classification and in the League the Belarusian players have a strong position. Maybe they compete in the next rounds between themselves and pursuers can still get a role in the race on the top position.
Tomorrow pairings of round 6, which round will be played on 4th December or earlier.
Statistics: Posted by Piet Bouma — Sat Nov 21, 2020 11:49
De Dick Brunaschool in Waddinxveen doet mee met 40 DamZ! DamSets.
Statistics: Posted by Marcel Kosters — Fri Nov 20, 2020 23:38
- Patterns: Pioneered by Fabien's Scan, strongest programs now for 8x8 checkers and 10x10 draughts. Input = K indices ranged 1..3^N for patterns of N squares, only K valid for every position. Fast index computations (PEXT) + direct lookup of K weights. No layers on top (sigmoid for training).
- Raw board: Pioneered in backgammon in the 1990s, now by Jonathan's GuiNN_checkers. Slightly stronger than Cake, still weaker than Kingsrow for 8x8 checkers. Input = all Piece entries (both type and square). 3 fully connected layers on top. Requires Python float weights -> C++ int conversion + SIMD programming + incremental updates (not yet done by Jonathan) to be fast during game play.
- NNUE: Pioneered in Shogi programs, now in Stockfish, currently strongest programs for chess, Shogi. Input = all King (square) * Piece (both type and square) entries. 3 fully connected layers on top. Same C++ machinery as for the above entry required (all implemented in Shogi and Stockfish).
- CNN: Pioneered by AlphaZero, currently strongest for Go, formerly for chess, Shogi. No successful attempts for checkers/draughts AFAIK. Input = all Piece (both type and square) entries, but the expensive comes from 3x3 convolutions in 40-80 layers deep.
In Shogi, they abbreviate the inputs by terms like KP for NNUE (king times piece). They even had KPP (king-piece-piece combinations) prior to NNUE. One could call Raw board networks "P". For NNUE they also add both K and P inputs to the KP tables (re-using the non-zero entries, this is called "factorization"), making it more of a KP + K + P network.
For draughts, there is no unique king piece, so one could either try the "P" type of networks, or "PP" networks: all Piece (both type and square) * Piece (both type and square) entries. For draughts, PP would be about as expensive as KP in chess. It would require serious resources, but still 4 orders of magnitude less than A0 type CNNs.
A slightly cheaper version might be called "PN" networks: all Piece (both type and square) * Neighbor (both type and square) entries. So only the 4 neighboring squares get computed. This is only slightly more expensive than the "P" type networks, yet might offer a flexible form of Scan-like patterns (speculative!).
Statistics: Posted by Rein Halbersma — Fri Nov 20, 2020 22:02