Mate in 15

Discussion of anything and everything relating to chess playing software and machines.

Moderators: hgm, Rebel, chrisw

Sven
Posts: 4052
Joined: Thu May 15, 2008 9:57 pm
Location: Berlin, Germany
Full name: Sven Schüle

Re: Mate in 15

Post by Sven »

jwes wrote:I suspect that Naum sees it very quickly, not instantly. White can apparently force an exchange in a few ply and then see the mate with egtbs.
As shown above, White captures the queen at ply 7 after 1.Qe7+ Qg5 (else 2.Qe1+ and mate) 2.Qe4 Qg4 3.Qe3 since Black must sacrifice its queen now. This forced variation will be seen very quickly by most engines (maybe in iteration 4 or even earlier), and with an EGTB lookup at ply 8 the perfect solution is easy to obtain.

Sven
Top
User avatar
michiguel
Posts: 6401
Joined: Thu Mar 09, 2006 8:30 pm
Location: Chicago, Illinois, USA

Re: Mate in 15

Post by michiguel »

Sven Schüle wrote:
jwes wrote:I suspect that Naum sees it very quickly, not instantly. White can apparently force an exchange in a few ply and then see the mate with egtbs.
As shown above, White captures the queen at ply 7 after 1.Qe7+ Qg5 (else 2.Qe1+ and mate) 2.Qe4 Qg4 3.Qe3 since Black must sacrifice its queen now. This forced variation will be seen very quickly by most engines (maybe in iteration 4 or even earlier), and with an EGTB lookup at ply 8 the perfect solution is easy to obtain.

Sven
Yes. This is Gaviota's output with 5 piece TBs. It sees the mate in 0.1 s

Code: Select all

        73   1:      0.0    -2.61  1.Qe7+ Qg5
       334   2:      0.0    -2.53  1.Qe7+ Qg5 2.Qd6
       451   3       0.0      :-)  1.Qe7+
      1157   3:      0.0    +0.00  1.Qe7+ Qg5 2.Qe4+ Qg4 3.Qe7+
      1659   4       0.0      :-)  1.Qe7+
      1844   4       0.0      :-)  1.Qe7+
      2029   4       0.0      :-)  1.Qe7+
      3451   4:      0.0    +5.16  1.Qe7+ Qg5 2.Qe4+ Qg4 3.Qe3 Qf3 4.Qxf3
      3664   5       0.0      :-)  1.Qe7+
      7996   5:      0.1  +Mat_12  1.Qe7+ Qg5 2.Qe4+ Qg4 3.Qe3 Qg2+ 4.Kxg2
                                   <=TABLE
     19377   6       0.1  +Mat_12  1.Qe7+ Qg5 2.Qe4+ Qg4 3.Qe3 Qg2+ 4.Kxg2
                                   <=TABLE
     22412   6&#58;      0.2  +Mat_12  1.Qe7+ Qg5 2.Qe4+ Qg4 3.Qe3 Qg2+ 4.Kxg2
                                   <=TABLE
     54300   7       0.3  +Mat_12  1.Qe7+ Qg5 2.Qe4+ Qg4 3.Qe3 Qg2+ 4.Kxg2
                                   <=TABLE
     58474   7&#58;      0.3  +Mat_12  1.Qe7+ Qg5 2.Qe4+ Qg4 3.Qe3 Qg2+ 4.Kxg2
                                   <=TABLE
    445601   8       1.2  +Mat_12  1.Qe7+ Qg5 2.Qe4+ Qg4 3.Qe3 Qg2+ 4.Kxg2
                                   <=TABLE
    455222   8&#58;      1.2  +Mat_12  1.Qe7+ Qg5 2.Qe4+ Qg4 3.Qe3 Qg2+ 4.Kxg2
                                   <=TABLE
   1161762   9       2.7  +Mat_12  1.Qe7+ Qg5 2.Qe4+ Qg4 3.Qe3 Qg2+ 4.Kxg2
                                   <=TABLE
   1189922   9&#58;      2.8  +Mat_12  1.Qe7+ Qg5 2.Qe4+ Qg4 3.Qe3 Qg2+ 4.Kxg2
                                   <=TABLE
   3201517  10       6.9  +Mat_12  1.Qe7+ Qg5 2.Qe4+ Qg4 3.Qe3 Qg2+ 4.Kxg2
                                   <=TABLE
   3297619  10&#58;      7.1  +Mat_12  1.Qe7+ Qg5 2.Qe4+ Qg4 3.Qe3 Qg2+ 4.Kxg2
                                   <=TABLE
   9888741  11      20.9  +Mat_12  1.Qe7+ Qg5 2.Qe4+ Qg4 3.Qe3 Qg2+ 4.Kxg2
                                   <=TABLE
  10182186  11&#58;     21.4  +Mat_12  1.Qe7+ Qg5 2.Qe4+ Qg4 3.Qe3 Qg2+ 4.Kxg2
                                   <=TABLE
Top
Sven
Posts: 4052
Joined: Thu May 15, 2008 9:57 pm
Location: Berlin, Germany
Full name: Sven Schüle

Re: Mate in 15

Post by Sven »

rightrook wrote:This is an interesting end game study!

[D]8/7Q/6p1/7p/6qk/8/7K/8 w - - 0 1

See how fast your engine can ann. a mate here....:-)
Thanks to that test position I also found a bug in my hash table/mate scoring code which caused my new engine to display the correct PV with a wrong score :-)

Sven
Top
Jan Brouwer
Posts: 201
Joined: Thu Mar 22, 2007 7:12 pm
Location: Netherlands

Re: Mate in 15 - update

Post by Jan Brouwer »

Jan Brouwer wrote:
zullil wrote:
Jan Brouwer wrote:
rightrook wrote:This is an interesting end game study!

[D]8/7Q/6p1/7p/6qk/8/7K/8 w - - 0 1

See how fast your engine can ann. a mate here....:-)
Rotor claims mate in 11:

info depth 17 seldepth 30 nodes 14333149 nps 1213953 score mate 11 hashfull 435 pv h7e7 g4g5 e7e4 g5g4 e4e3 g4e2
[D] 8/8/6p1/7p/6qk/4Q3/7K/8 b - - 5 3
What if Black plays g4g2 here instead of g4e2?
Rotor then claims mate in 9. Hmm...
Found the bug!

It took some effort, the first tree dump was so large that trying to view it caused my poor old laptop to grind to a halt...
Rotor uses two bounds in the transposition table (one day I will have to test if this actually helps).
When one of the bounds is a mate score and the other is not, then the mate score bound is not adjusted from node-relative to root-relative on retrieval, but may still be used.
This bug has been there since version 0.1!

Thanks to Robert Basham for posting this position and to Louis Zulli for noticing something fishy :)
Top
User avatar
AdminX
Posts: 6340
Joined: Mon Mar 13, 2006 2:34 pm
Location: Acworth, GA

Re: Mate in 15

Post by AdminX »

Analysis by Critter 0.60 64-bit:

1.Qe7+ Qg5 2.Qe4+ Qg4 3.Qe3 Qg2+ 4.Kxg2 Kg4 5.Qf3+ Kg5 6.Kh3 Kh6 7.Qf6 Kh7 8.Kh4 g5+ 9.Kxg5 h4 10.Qh6+ Kg8 11.Kg6 h3 12.Qg7#
= (0.00) Depth: 6/10 00:00:00 1kN
+- (#12) Depth: 25/36 00:00:06 22167kN
"Good decisions come from experience, and experience comes from bad decisions."
__________________________________________________________________
Ted Summers
Top
Edmund
Posts: 670
Joined: Mon Dec 03, 2007 3:01 pm
Location: Barcelona, Spain

Re: Mate in 15 - update

Post by Edmund »

Jan Brouwer wrote: Found the bug!

It took some effort, the first tree dump was so large that trying to view it caused my poor old laptop to grind to a halt...
Rotor uses two bounds in the transposition table (one day I will have to test if this actually helps).
When one of the bounds is a mate score and the other is not, then the mate score bound is not adjusted from node-relative to root-relative on retrieval, but may still be used.
This bug has been there since version 0.1!

Thanks to Robert Basham for posting this position and to Louis Zulli for noticing something fishy :)
I just happend to run a test like this myself. Glass selfplay 700 games at 1 minutes / game

The Version storing 2 bounds was weaker:

Code: Select all

ID  Date / Time  	  V1  	V2  	TC  	Games  	Score  	LOS  	ELO  	Remarks  	Conclusions
58	2010-04-14 19&#58;46 	mod 	rec 	60s 	715 	47.55% 	5.63% 	-17.02 	extended TT 	failed
However I wouldn't be surprised if it was completly different under another setting.
Top
metax
Posts: 344
Joined: Wed Sep 23, 2009 5:56 pm
Location: Germany

Re: Mate in 15 - update

Post by metax »

Strange... My engine has absolutely no clue of what is going on here... may be a null-move problem with the zugzwang. I'll have to take a closer look at this. :?


[D]8/7Q/6p1/7p/6qk/8/7K/8 w - -

Engine: ChessMind 0.72 (0 MB)
by metax

15/41 0:01 0.00 1.Qe7+ Qg5 2.Qb4+ Qg4 3.Qe7+ (421.275) 317

16/46 0:02 0.00 1.Qe7+ Qg5 2.Qb4+ Qg4 3.Qe7+ (763.343) 334

17/47 0:02 0.00 1.Qe7+ Qg5 2.Qb4+ Qg4 3.Qe7+ (989.902) 342

18/52 0:03 0.00 1.Qe7+ Qg5 2.Qb4+ Qg4 3.Qe7+ (1.328.544) 351

19/55 0:04 0.00 1.Qe7+ Qg5 2.Qb4+ Qg4 3.Qe7+ (1.757.786) 357

20/58 0:06 0.00 1.Qe7+ Qg5 2.Qb4+ Qg4 3.Qe7+ (2.356.749) 364

21/58 0:09 0.00 1.Qe7+ Qg5 2.Qb4+ Qg4 3.Qe7+ (3.564.449) 387

22/64 0:19 0.00 1.Qe7+ Qg5 2.Qb4+ Qg4 3.Qe7+ (8.092.356) 405

23/65 0:47 0.00 1.Qe7+ Qg5 2.Qb4+ Qg4 3.Qe7+ (21.697.180) 452

24/66 1:08 0.00 1.Qe7+ Qg5 2.Qb4+ Qg4 3.Qe7+ (30.997.002) 455

best move: Qh7-e7 time: 1:12.213 min n/s: 452.000 nodes: 32.130.557
Top
metax
Posts: 344
Joined: Wed Sep 23, 2009 5:56 pm
Location: Germany

Re: Mate in 15 - update

Post by metax »

ChessMind with null-move turned off finds it...


[D]8/7Q/6p1/7p/6qk/8/7K/8 w - -

Engine: ChessMind 0.72 (0 MB)
by metax

10/27 0:01 +8.75 1.Qe7+ Qg5 2.Qe4+ Qg4 3.Qe3 Qh3+
4.Qxh3+ Kg5 5.Qe3+ Kf6 6.Qd4+ Ke6
7.Qf2 Ke7 8.Qc5+ (427.056) 314

11/29 0:02 +8.89 1.Qe7+ Qg5 2.Qe4+ Qg4 3.Qe3 Qh3+
4.Qxh3+ Kg5 5.Qe3+ Kf6 6.Qf4+ Ke6
7.Kg3 Ke7 8.Kf2 Ke8 (774.723) 352

12/30 0:04 +9.07 1.Qe7+ Qg5 2.Qe4+ Qg4 3.Qe3 Qf3
4.Qxf3 Kg5 5.Kg3 Kh6 6.Qc3 Kg5
7.Qe3+ (1.552.874) 375

13/34 0:08 +10.83 1.Qe7+ Qg5 2.Qe4+ Qg4 3.Qe3 Qf3
4.Qxf3 Kg5 5.Kg3 h4+ 6.Kh3 Kh6
7.Kxh4 g5+ 8.Kg4 (3.457.038) 395

14/37 0:15 +10.83 1.Qe7+ Qg5 2.Qe4+ Qg4 3.Qe3 Qf3
4.Qxf3 Kg5 5.Kg3 h4+ 6.Kh3 Kh6
7.Kxh4 g5+ 8.Kg4 Kg6 9.Qf5+ (6.047.834) 398

15/40 0:46 +11.06 1.Qe7+ Qg5 2.Qe4+ Qg4 3.Qe3 Qf3
4.Qxf3 Kg5 5.Kg3 h4+ 6.Kh3 Kh6
7.Kxh4 g5+ 8.Kg4 Kg6 9.Qf5+ Kh6
10.Qxg5+ (20.210.369) 430

16/42 1:40 +11.32 1.Qe7+ Qg5 2.Qe4+ Qg4 3.Qe3 Qf3
4.Qxf3 Kg5 5.Kg3 h4+ 6.Kh3 Kh6
7.Kxh4 g5+ 8.Kg4 Kg6 9.Qf5+ Kh6
10.Qxg5+ Kh7 (44.282.077) 440
Top
F. Bluemers
Posts: 868
Joined: Thu Mar 09, 2006 11:21 pm
Location: Nederland

Re: Mate in 15

Post by F. Bluemers »

Dirty

Code: Select all

18	+319.75	4.8M	0&#58;02.48	Qe7+ Qg5 Qe4+ Qg4 Qe3 Qg2+ Kxg2 Kg4 Qf3+ Kg5 Kh3 Kh6 Kg3 Kg7 Kh4 Kg8 Qf6 g5+ Kxg5 Kh7 Qf7+ Kh8 Kh6 h4 Qf8#
 17	+319.75	4.4M	0&#58;02.19	Qe7+ Qg5 Qe4+ Qg4 Qe3 Qg2+ Kxg2 Kg4 Qf3+ Kg5 Kh3 h4 Qf8 Kh5 Qf6 g5 Qf5 Kh6 Qf7 g4+ Kxg4 h3 Kf5 h2 Qg6#
 17	+319.75	4.1M	0&#58;02.05	Qe7+ Qg5 Qe4+ Qg4 Qe3 Qg2+ Kxg2 Kg4 Qf3+ Kg5 Kh3 h4 Qf8 Kh5 Qf6 g5 Qf5 Kh6 Qf7 g4+ Kxg4 h3 Kf5 h2 Qg6#
 16	+13.34	2.5M	0&#58;01.37	Qe7+ Qg5 Qe4+ Qg4 Qe3 Qe2+ Qxe2 Kg5 Qf3 Kh6 Kg3 Kg7 Kh4 Kg8 Qf6 g5+ Kxh5 Kh7 Kg4 Kg8 Kxg5 Kh7 Qg6+ Kh8 Qh5+ Kg7 Qf3 Kh7 Qf7+ Kh8 Qf8+ Kh7 Qf6 Kg8 Kf5 Kh7 Qf7+ Kh8
 16	+13.34	2.3M	0&#58;01.29	Qe7+ Qg5 Qe4+ Qg4 Qe3 Qe2+ Qxe2 Kg5 Qf3 Kh6 Kg3 Kg7 Kh4 Kg8 Qf6 g5+ Kxh5 Kh7 Kg4 Kg8 Kxg5 Kh7 Qg6+ Kh8 Qh5+ Kg7 Qf3 Kh7 Qf7+ Kh8 Qf8+ Kh7 Qf6 Kg8 Kf5 Kh7 Qf7+ Kh8
 15	+13.07	1.5M	0&#58;00.87	Qe7+ Qg5 Qe4+ Qg4 Qe3 Qe2+ Qxe2 Kg5 Qf3 Kh6 Kg3 Kg7 Kh4 Kg8 Qf6 g5+ Kxh5 g4 Kxg4 Kh7 Kf4
 15	+13.07	1.4M	0&#58;00.82	Qe7+ Qg5 Qe4+ Qg4 Qe3 Qe2+ Qxe2 Kg5 Qf3 Kh6 Kg3 Kg7 Kh4 Kg8 Qf6 g5+ Kxh5 g4 Kxg4 Kh7 Kf4
 14	+12.98	896590	0&#58;00.59	Qe7+ Qg5 Qe4+ Qg4 Qe3 Qe2+ Qxe2 Kg5 Qf3 Kh6 Qf8+ Kh7 Qf7+ Kh6 Kg3 g5 Qf6+ Kh7 Qxg5 Kh8 Qh6+ Kg8
 14	+12.98	816854	0&#58;00.54	Qe7+ Qg5 Qe4+ Qg4 Qe3 Qe2+ Qxe2 Kg5 Qf3 Kh6 Qf8+ Kh7 Qf7+ Kh6 Kg3 g5 Qf6+ Kh7 Qxg5 Kh8 Qh6+ Kg8
 13	+12.88	603508	0&#58;00.43	Qe7+ Qg5 Qe4+ Qg4 Qe3 Qe2+ Qxe2 Kg5 Qf3 Kh6 Qf6 Kh7 Kg3 h4+ Kxh4 Kg8 Qxg6+ Kf8
 13	+12.88	546149	0&#58;00.40	Qe7+ Qg5 Qe4+ Qg4 Qe3 Qe2+ Qxe2 Kg5 Qf3 Kh6 Qf6 Kh7 Kg3 h4+ Kxh4 Kg8 Qxg6+ Kf8
 12	+12.43	426210	0&#58;00.34	Qe7+ Qg5 Qe4+ Qg4 Qe3 Qe2+ Qxe2 Kg5 Qe5+ Kg4 Qf6 h4 Qxg6+ Kf4 Kg2 Ke5 Qh5+ Kf6
 12	+12.43	382078	0&#58;00.32	Qe7+ Qg5 Qe4+ Qg4 Qe3 Qe2+ Qxe2 Kg5 Qe5+ Kg4 Qf6 h4 Qxg6+ Kf4 Kg2 Ke5 Qh5+ Kf6
 11	+11.40	221099	0&#58;00.20	Qe7+ Qg5 Qe4+ Qg4 Qe3 Qe2+ Qxe2 Kg5 Qf3 Kh6 Qf8+ Kg5 Kg3 h4+ Kh3 Kh5 Qh8+ Kg5 Qe5+ Kh6
 11	+11.40	188155	0&#58;00.17	Qe7+ Qg5 Qe4+ Qg4 Qe3 Qe2+ Qxe2 Kg5 Qf3 Kh6 Qf8+ Kg5 Kg3 h4+ Kh3 Kh5 Qh8+ Kg5 Qe5+ Kh6
 10	+10.25	144551	0&#58;00.14	Qe7+ Qg5 Qe4+ Qg4 Qe3 Qe2+ Qxe2 Kg5 Qf3 Kh6 Qf8+ Kh7 Kg3 g5
 10	+10.25	126323	0&#58;00.12	Qe7+ Qg5 Qe4+ Qg4 Qe3 Qe2+ Qxe2 Kg5 Qf3 Kh6 Qf8+ Kh7 Kg3 g5
  9	+9.57	94563	0&#58;00.07	Qe7+ Qg5 Qe4+ Qg4 Qe3 Qe2+ Qxe2 Kg5 Kg3 Kf6 Kf4 g5+ Kg3
  9	+9.57	76416	0&#58;00.07	Qe7+ Qg5 Qe4+ Qg4 Qe3 Qe2+ Qxe2 Kg5 Kg3 Kf6 Kf4 g5+ Kg3
  8	+9.21	53077	0&#58;00.04	Qe7+ Qg5 Qe4+ Qg4 Qe3 Qe2+ Qxe2 Kg5 Kg3 Kf5 Kh4
  8	+9.21	40991	0&#58;00.04	Qe7+ Qg5 Qe4+ Qg4 Qe3 Qe2+ Qxe2 Kg5 Kg3 Kf5 Kh4
  7	+8.84	27571	0&#58;00.03	Qe7+ Qg5 Qe4+ Qg4 Qe3 Qf3 Qxf3 Kg5 Kg3
  7	+8.84	20673	0&#58;00.03	Qe7+ Qg5 Qe4+ Qg4 Qe3 Qf3 Qxf3 Kg5 Kg3
  6	+8.42	14754	0&#58;00.01	Qe7+ Qg5 Qe4+ Qg4 Qe3 Qf3 Qxf3 Kg5
  6	+8.42	9966	0&#58;00.01	Qe7+ Qg5 Qe4+ Qg4 Qe3 Qf3 Qxf3 Kg5
  5	0.00	5917	0&#58;00.01	Qe7+ Qg5 Qe4+ Qg4 Qe7+
  5	0.00	4203	0&#58;00.01	Qe7+ Qg5 Qe4+ Qg4 Qe7+
  4	0.00	3057	0&#58;00.00	Qe7+ Qg5 Qe4+ Qg4 Qe7+
  4	0.00	1716	0&#58;00.00	Qe7+ Qg5 Qe4+ Qg4 Qe7+
  3	0.00	1062	0&#58;00.00	Qe7+ Qg5 Qe4+ Qg4 Qe7+
  3	0.00	677	0&#58;00.00	Qe7+ Qg5 Qe4+ Qg4 Qe7+
  2	-3.62	331	0&#58;00.00	Qe7+ Qg5 Qe1+ Kg4
  2	-3.62	272	0&#58;00.00	Qe7+ Qg5 Qe1+ Kg4
  1	-4.39	79	0&#58;00.00	Qe7+ Qg5
Reminds me of the stalemates:http://www.talkchess.com/forum/viewtopi ... 57&t=33782 :wink:
Top

Return to “Computer Chess Club: General Topics”