Checkmate evaluation test

Discussion of chess software programming and technical issues.

Moderator: Ras

Chessnut1071
Posts: 313
Joined: Tue Aug 03, 2021 2:41 pm
Full name: Bill Beame

Checkmate evaluation test

Post by Chessnut1071 »

[fen]3kNr2/p1ppppNq/Ppr1p2p/8/1R6/8/1Q3BB1/1R2b2K w - - 0 1[/fen]

The author claims the above puzzle can be solved in 7 moves, however, he doesn't supply the solution. My engine requires 10 moves, 121,550,873,524 nodes, and over 3 hours to solve.

If you have a more efficient solution, I would like to see number of moves, node count, time and your evaluation function.

My evaluation function includes: 1. check 2. history 3. sliders 4. pawn promotion 5. captures 6. enemy king moves 7. distance to enemy king in weighted order of priority.

engine: Intel i7 2.6 GHz C# Bitboard PEXT
Joost Buijs
Posts: 1635
Joined: Thu Jul 16, 2009 10:47 am
Location: Almere, The Netherlands

Re: Checkmate evaluation test

Post by Joost Buijs »

For what it is worth. My engine uses NNUE for evaluation, it finds a mate in 11 at depth 18 in 121 seconds, after 159 seconds it sees a mate in 10 at depth 19 and it sticks to this. So I'm quite sure that this position is a mate in 10 and not a mate in 7. The node count at depth 19 is 3611394178, my engine used 16 threads on a i9-10980XE.
Chessnut1071
Posts: 313
Joined: Tue Aug 03, 2021 2:41 pm
Full name: Bill Beame

Re: Checkmate evaluation test

Post by Chessnut1071 »

Joost Buijs wrote: Sun Oct 16, 2022 3:53 pm For what it is worth. My engine uses NNUE for evaluation, it finds a mate in 11 at depth 18 in 121 seconds, after 159 seconds it sees a mate in 10 at depth 19 and it sticks to this. So I'm quite sure that this position is a mate in 10 and not a mate in 7. The node count at depth 19 is 3611394178, my engine used 16 threads on a i9-10980XE.
thx. I found 1. Rxp lead to mate in 10. I seriously doubt there is a 7-move mate here. How many cores in your i9? I'm limited to 1 core and 1 thread.
Joost Buijs
Posts: 1635
Joined: Thu Jul 16, 2009 10:47 am
Location: Almere, The Netherlands

Re: Checkmate evaluation test

Post by Joost Buijs »

Chessnut1071 wrote: Sun Oct 16, 2022 4:06 pm
Joost Buijs wrote: Sun Oct 16, 2022 3:53 pm For what it is worth. My engine uses NNUE for evaluation, it finds a mate in 11 at depth 18 in 121 seconds, after 159 seconds it sees a mate in 10 at depth 19 and it sticks to this. So I'm quite sure that this position is a mate in 10 and not a mate in 7. The node count at depth 19 is 3611394178, my engine used 16 threads on a i9-10980XE.
thx. I found 1. Rxp lead to mate in 10. I seriously doubt there is a 7-move mate here. How many cores in your i9? I'm limited to 1 core and 1 thread.
The i9-10980XE has 18 cores, I used 16, it runs at the default clock of 3.0 GHz.
Uri Blass
Posts: 10799
Joined: Thu Mar 09, 2006 12:37 am
Location: Tel-Aviv Israel

Re: Checkmate evaluation test

Post by Uri Blass »

Chessnut1071 wrote: Sun Oct 16, 2022 2:34 pm [fen]3kNr2/p1ppppNq/Ppr1p2p/8/1R6/8/1Q3BB1/1R2b2K w - - 0 1[/fen]

The author claims the above puzzle can be solved in 7 moves, however, he doesn't supply the solution. My engine requires 10 moves, 121,550,873,524 nodes, and over 3 hours to solve.

If you have a more efficient solution, I would like to see number of moves, node count, time and your evaluation function.

My evaluation function includes: 1. check 2. history 3. sliders 4. pawn promotion 5. captures 6. enemy king moves 7. distance to enemy king in weighted order of priority.

engine: Intel i7 2.6 GHz C# Bitboard PEXT
Chest prove no mate in 7 and no mate in 8 with one core in a few minutes so the author is clearly wrong.
Uri Blass
Posts: 10799
Joined: Thu Mar 09, 2006 12:37 am
Location: Tel-Aviv Israel

Re: Checkmate evaluation test

Post by Uri Blass »

chest proved also no mate in 9 so it is mate in 10.
Chessnut1071
Posts: 313
Joined: Tue Aug 03, 2021 2:41 pm
Full name: Bill Beame

Re: Checkmate evaluation test

Post by Chessnut1071 »

Uri Blass wrote: Sun Oct 16, 2022 6:47 pm chest proved also no mate in 9 so it is mate in 10.
Hope it's 1. RxP. How long did it take? Glad you confirmed it, that's too deep for me to debug my engine quickly, I could have wasted more than a week.
Uri Blass
Posts: 10799
Joined: Thu Mar 09, 2006 12:37 am
Location: Tel-Aviv Israel

Re: Checkmate evaluation test

Post by Uri Blass »

Chessnut1071 wrote: Sun Oct 16, 2022 7:09 pm
Uri Blass wrote: Sun Oct 16, 2022 6:47 pm chest proved also no mate in 9 so it is mate in 10.
Hope it's 1. RxP. How long did it take? Glad you confirmed it, that's too deep for me to debug my engine quickly, I could have wasted more than a week.
I confirmed no mate in 9 with chest in 2 hours or something like it.
I stopped chest and slowchess found mate in 10 starts with Rxp in few seconds but slowchess is not designed to prove the fastest mate so I trust it that it is mate in 10 and I believe no mate in 9 based on chest.
JVMerlino
Posts: 1398
Joined: Wed Mar 08, 2006 10:15 pm
Location: San Francisco, California

Re: Checkmate evaluation test

Post by JVMerlino »

The King from CM9000 needed 3 seconds to report that Rxb6 is also a Mate in 10.

Code: Select all

Time	Depth	Score	Positions	Moves
0:00	1/3	3.88	1017		1.Be4 f5 2.Bxc6 Bxb4 3.Qxb4 dxc6
0:00	1/3	6.03	1429		1.Rxe1 Rc2 2.Bg3 d6
0:00	1/3	6.05	1875		1.Bxe1 Rc2 2.Be4
0:00	1/4	6.71	4379		1.Bxe1 Rc5 2.Be4 f5 3.Bg2
0:00	1/5	6.88	19347		1.Bxe1 d5 2.Rc1 Rxc1 3.Qxc1 c5 4.Qf4
0:00	1/6	7.38	82784		1.Bxe1 Rc2 2.Be4 Rxb2 3.Bxh7 Rxb4 4.Rxb4 Rxe8 5.Nxe8 Kxe8
0:00	1/6	7.64	114960		1.Rxe1 Rc2 2.Qd4 Rxe8 3.Be4 Qxg7 4.Qxg7 Rxf2 5.Qxh6
0:00	1/7	8.50	299095		1.Rxe1 Rc2 2.Rxb6 Rxb2 3.Rxb2 d5 4.Rc1 Rxe8 5.Rb8+ Kd7 6.Nxe8 c6 7.Bxa7 e5 8.Bb6
0:01	1/8	9.00	1146901		1.Rxe1 Qc2 2.Bxc6 Qxc6+ 3.Kg1 d6 4.Qc1 Qd7 5.Qxh6 c5
0:01	1/8	Mate13	1620991		1.Rxb6 cxb6 2.Bxb6+ Rxb6 3.Qxb6+ axb6 4.a7 Qc2 5.Rxb6 d5 6.a8=Q+ Qc8 7.Rb8 Kd7 8.Qa7+ Kc6 9.Qa4+ Kc5 10.Rxc8+ Kb6 11.Rc6+ Kb7 12.Qa6+ Kb8 13.Rc8#
0:02	1/9	Mate12	2636101		1.Rxb6 cxb6 2.Bxb6+ Rxb6 3.Qxb6+ axb6 4.a7 Qc2 5.Rxb6 d5 6.a8=Q+ Qc8 7.Rb8 Kd7 8.Qa4+ Kd8 9.Rxc8+ Kxc8 10.Qa8+ Kd7 11.Qb7+ Kd8 12.Qc7#
0:03	1/10	Mate10	4781759		1.Rxb6 cxb6 2.Bxb6+ Rxb6 3.Qxb6+ axb6 4.a7 Qc2 5.Rxb6 Qc8 6.Rb8 Bg3 7.Rxc8+ Kxc8 8.a8=Q+ Bb8 9.Qb7+ Kd8 10.Qxb8#
JVMerlino
Posts: 1398
Joined: Wed Mar 08, 2006 10:15 pm
Location: San Francisco, California

Re: Checkmate evaluation test

Post by JVMerlino »

My engine, rated about 2600, with 8 cores, needed 4 minutes and 26 seconds to announce Mate in 10 with Rxb6, searching almost 7.9B nodes.

21 32748 44598 7,881,802,951 b4b6 c7b6 (17672 KNPS)(Mate in 10)