Mate in 82?

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

Moderators: bob, hgm, Harvey Williamson

Forum rules
This textbox is used to restore diagrams posted with the [d] tag before the upgrade.
zullil
Posts: 5606
Joined: Mon Jan 08, 2007 11:31 pm
Location: PA USA
Full name: Louis Zulli

Re: Mate in 82?

Post by zullil » Thu Apr 04, 2019 6:52 pm

Steve Maughan wrote:
Thu Apr 04, 2019 6:02 pm
If anyone is interested, here's the mate in 82 that Komodo 12.3 found:

Was this with default settings? In particular, was null move pruning disabled?

Dann Corbit
Posts: 9993
Joined: Wed Mar 08, 2006 7:57 pm
Location: Redmond, WA USA
Contact:

Re: Mate in 82?

Post by Dann Corbit » Thu Apr 04, 2019 7:10 pm

I just ran it myself and default settings solved mate in 82 in 1:09 at depth 47 with only 11 threads.

Code: Select all

FEN: 8/1p1p2p1/5p1p/5P1P/1p1p1P2/pP1P1P2/P1k1B3/4K3 w - - 0 1 

Komodo-12.3-64bit:
  10+	00:00	     133,505	16,136,165	-0.36	Bf1 Kb2 Kd2
  10+	00:00	     134,810	15,653,622	-0.15	Bf1 Kb2 Kd2
  10+	00:00	     137,073	15,057,729	 0.00	Bf1 Kb2 Kd2
  10	00:00	     149,565	14,453,408	 0.00	Bf1 Kb2 Kd2 Kxa2 Kc2 Ka1 Bh3 d6 Bf1 b5 Bh3 Ka2 Bg2 Ka1 Bh3
  11	00:00	     157,823	14,772,448	 0.00	Bf1 Kb2 Kd2 Kb1 Bg2 b6 Bh3 Kxa2 Kc2 Ka1 Bg4 Ka2 Bh3
  12	00:00	     173,067	14,936,441	 0.00	Bf1 Kb1 Kd2 Kxa2 Kc2 Ka1 Bh3 b6 Bg4 Ka2 Bh3 Ka1
  13	00:00	   1,179,885	27,278,820	 0.00	Bf1 Kb1 Kd2 Kxa2 Kc2 Ka1 Bh3 b6 Bg4 Ka2 Bh3 Ka1
  14	00:00	   1,186,582	27,139,993	 0.00	Bf1 Kb1 Kd2 Kxa2 Kc2 Ka1 Bh3 b6 Bg4 Ka2 Bh3 Ka1
  15	00:00	   1,204,592	25,123,096	 0.00	Bf1 Kb1 Kd2 Kxa2 Kc2 Ka1 Bh3 Ka2 Bg4 Ka1 Bh3
  16	00:00	   1,241,605	24,512,605	 0.00	Bf1 Kb1 Kd2 Kxa2 Kc2 Ka1 Bh3 Ka2 Bg4 Ka1 Bh3
  17	00:00	   1,286,067	23,748,243	 0.00	Bf1 Kb1 Kd2 Kxa2 Kc2 Ka1 Bh3 Ka2 Bg4 Ka1 Bh3
  18	00:00	   1,298,014	23,369,207	 0.00	Bf1 Kb1 Kd2 Kxa2 Kc2 Ka1 Bh3 Ka2 Bg4 Ka1 Bh3
  19	00:00	   1,746,754	24,342,249	 0.00	Bf1 Kb1 Kd2 Kxa2 Kc2 Ka1 Bh3 Ka2 Bg4 Ka1 Bh3
  20	00:00	   2,014,828	24,889,741	 0.00	Bf1 Kb1 Kd2 Kxa2 Kc2 Ka1 Bh3 Ka2 Bg4 Ka1 Bh3
  21	00:00	   2,074,510	23,989,591	 0.00	Bf1 Kb1 Kd2 Kxa2 Kc2 Ka1 Bh3 Ka2 Bg4 Ka1 Bh3
  22	00:00	   4,843,266	28,935,416	 0.00	Bf1 Kb1 Kd2 Kxa2 Kc2 Ka1 Bh3 Ka2 Bg4 Ka1 Bh3
  23	00:00	   4,862,084	28,505,120	 0.00	Bf1 Kb1 Kd2 Kxa2 Kc2 Ka1 Bh3 Ka2 Bg4 Ka1 Bh3
  24	00:00	   4,987,106	27,844,918	 0.00	Bf1 Kb1 Kd2 Kxa2 Kc2 Ka1 Bh3 Ka2 Bg4 Ka1 Bh3
  25	00:00	   5,306,227	27,254,234	 0.00	Bf1 Kb1 Kd2 Kxa2 Kc2 Ka1 Bh3 Ka2 Bg4 Ka1 Bh3
  26	00:00	   5,433,340	26,678,668	 0.00	Bf1 Kb1 Kd2 Kxa2 Kc2 Ka1 Bh3 Ka2 Bg4 Ka1 Bh3
  27	00:00	   6,036,844	26,326,197	 0.00	Bf1 Kb1 Kd2 Kxa2 Kc2 Ka1 Bh3 Ka2 Bg4 Ka1 Bh3
  28	00:00	   6,691,487	24,491,917	 0.00	Bf1 Kb1 Kd2 Kb2 Kd1 Kb1 Bh3 Kxa2 Kc2 Ka1 Bg2 Ka2 Bh3
  29+	00:00	   7,188,014	16,836,576	+0.06	Bf1 Kb1 Kd2
  29-	00:00	   7,340,628	14,406,221	+0.02	Bf1 Kc1
  29	00:00	   9,613,800	12,492,344	 0.00	Bf1 Kc1 Ke2 Kc2 Bg2 Kc3 Bf1 Kc2
  30+	00:00	   9,680,044	12,483,411	+0.06	Bf1 Kc1 Ke2
  30+	00:00	   9,686,107	12,478,475	+0.14	Bf1 Kc1 Ke2
  30+	00:00	   9,701,782	12,450,464	+0.25	Bf1 Kc1 Ke2
  30	00:00	   9,842,410	11,904,162	+0.37	Bf1 Kc1 Ke2 Kc2 Bg2 Kc3 Bh3 Kc2 Bf1 Kc1 Ke1 Kc2 Be2 Kc1 Bd1 Kb1 Kd2 Kxa2 Kc2 Ka1 Kc1 Ka2 Bc2 b6 Bd1 Ka1 Be2 Ka2 Kc2 Ka1 Bf1 Ka2 Bg2 Ka1 Bh3 Ka2 Bf1 Ka1 Be2 Ka2 Bd1 Ka1 Kc1 Ka2
  31+	00:01	  13,093,095	11,487,149	+0.43	Bf1 Kc1 Ke2
  31+	00:01	  13,100,602	11,479,724	+0.51	Bf1 Kc1 Ke2
  31+	00:01	  13,229,709	11,378,927	+0.62	Bf1 Kc1 Ke2
  31-	00:01	  13,405,996	11,010,114	+0.55	Bf1 Kc1
  31+	00:01	  15,970,634	9,431,695	+0.67	Bf1 b5
  31-	00:01	  16,308,640	9,414,230	+0.53	Bf1 Kc1
  31+	00:01	  16,365,247	9,415,714	+0.74	Bf1 Kc1 Bh3
  31+	00:01	  16,442,865	9,422,938	+1.46	Bf1 Kb1 Kd2
  31-	00:01	  16,448,891	9,424,472	+0.06	Bf1 Kc1
  31	00:01	  16,581,637	9,461,908	 0.00	Bf1 Kc1 Bh3 Kc2 Ke2 Kc1 Bg2 Kc2 Bf1 Kc1 Ke1 Kc2 Be2 Kc1 Bd1 Kb2 Kd2 Kb1 Be2 Kb2 Kd1 Kb1 Bf1 Kb2 Be2
  32	00:01	  17,193,612	9,036,470	 0.00	Bf1 Kc1 Bh3 Kc2 Ke2 Kc1 Bg2 Kc2 Bf1 Kc1 Ke1 Kc2 Be2 Kc1 Bd1 Kb2 Kd2 Kb1 Be2 Kb2 Kd1 Kb1 Bf1 Kb2 Be2
  33	00:02	  26,344,261	10,121,716	 0.00	Bf1 Kc1 Bh3 Kc2 Ke2 Kc3 Bg2 Kc2 Bf1 Kc1 Ke1 Kc2 Be2 Kb1 Kd1 Kxa2 Kc2 Ka1 Kc1 Ka2 Kc2
  34	00:03	  32,852,994	10,851,445	 0.00	Bf1 Kc1 Bh3 Kc2 Ke2 Kc3 Bg2 Kc2 Bf1 Kc1 Ke1 Kc2 Be2 Kb1 Kd1 Kxa2 Kc2 Ka1 Kc1 Ka2 Kc2
  35	00:03	  33,852,232	10,092,174	 0.00	Bf1 Kc1 Bh3 Kc2 Ke2 Kc3 Bg2 Kc2 Bf1 Kc1 Ke1 Kc2 Be2 Kb1 Kd1 Kxa2 Kc2 Ka1 Kc1 Ka2 Kc2
  36	00:03	  34,606,937	9,114,239	 0.00	Bf1 Kc1 Bh3 Kc2 Ke2 Kc3 Bg2 Kc2 Bf1 Kc1 Ke1 Kc2 Be2 Kb1 Kd1 Kxa2 Kc2 Ka1 Kc1 Ka2 Kc2
  37+	00:05	  37,057,437	6,828,952	+0.06	Bf1 Kc1 Ke2
  37+	00:06	  38,714,527	6,248,238	+0.14	Bf1 Kc1 Ke2
  37+	00:06	  38,885,832	6,224,485	+0.25	Bf1 Kb1 Kd2
  37+	00:06	  38,945,537	6,206,698	+0.40	Bf1 Kc1 Ke2
  37+	00:06	  39,033,644	6,191,469	+0.61	Bf1 Kc1 Ke2
  37-	00:06	  39,424,247	6,133,707	+0.47	Bf1 Kc1
  37+	00:07	  47,950,092	6,047,805	+0.67	Bf1 Kc1 Ke2
  37+	00:07	  48,015,862	6,051,333	+1.49	Bf1 g5
  37-	00:07	  48,033,913	6,051,483	+1.04	Bf1 Kc1
  37	00:07	  48,160,324	6,057,590	 0.00	Bf1 Kc1 Ke2 Kc2 Bg2 Kc1 Bh3 Kc2 Bf1 Kc1 Ke1 Kc2 Be2 Kb2 Kd2 Kxa2 Kc2 Ka1 Bd1 Ka2 Be2
  38	00:08	  50,333,602	5,946,905	 0.00	Bf1 Kb2 Kd2 Kb1 Bh3 Kb2 Kd1 Kb1 Bg2 Kb2 Kd2 Kb1 Bh3
  39	00:08	  51,329,995	5,822,147	 0.00	Bf1 Kc1 Bh3 Kc2 Ke2 Kb2 Kd2 Kb1 Bf1 Kb2 Bg2 Kb1 Bh3 Kb2 Kd1 Kc3 Kc1 Kxd3 Bf1+ Kc3 Bc4 d3 Bd5 d2+ Kd1 Kb2 Kxd2 Kxa2 Kc2 b6 Bc4 Ka1 Bd3 d5 Bb5 d4 Bd3 Ka2 Bb5 Ka1
  40	00:09	  59,214,319	6,161,262	 0.00	Bf1 Kc1 Bh3 Kc2 Ke2 Kb2 Kd2 Kb1 Bf1 Kb2 Bg2 Kb1 Bh3 Kb2 Kd1 Kc3 Kc1 Kxd3 Bf1+ Kc3 Bc4 d3 Bd5 d2+ Kd1 Kb2 Kxd2 Kxa2 Kc2 b6 Bc4 Ka1 Bd3 d5 Bb5 d4 Bd3 Ka2 Bf1 Ka1 Bd3
  41	00:10	  71,486,441	6,724,751	 0.00	Bf1 Kc1 Bh3 Kc2 Ke2 Kc1 Bg2 Kc2 Bf1 Kc3 Kd1 Kb2 Bg2 Kc3 Bf1
  42+	00:10	  73,316,025	6,674,526	+0.06	Bf1 Kc1 Bh3
  42+	00:11	  73,548,944	6,664,615	+0.14	Bf1 Kc1 Bh3
  42+	00:11	  74,622,759	6,357,043	+0.73	Bf1 Kc1 Bh3
  42+	00:11	  74,631,859	6,356,610	+1.03	Bf1 Kb1 Kd2
  42+	00:11	  74,662,703	6,355,171	+1.44	Bf1 Kb1 Kd2
  42-	00:12	  79,336,923	6,448,068	+1.16	Bf1 Kc1
  42-	00:12	  84,472,604	6,527,075	+0.38	Bf1 Kc1
  42	00:12	  84,512,213	6,527,794	 0.00	Bf1 Kc1 Bh3 Kc2 Ke2 Kc1 Bg2 Kc2 Bf1 Kc1 Bg2
  43+	00:13	  85,678,819	6,539,511	+0.06	Bf1 Kc1 Bh3
  43+	00:13	  86,185,185	6,547,067	+0.14	Bf1 Kc1 Bh3
  43+	00:13	  93,309,525	6,686,186	+1.86	Bf1 Kc1 Bh3
  43-	00:16	 107,410,349	6,487,220	+1.47	Bf1 Kc1
  43+	00:16	 109,920,953	6,567,930	+2.01	Bf1 Kc1 Bg2
  43	00:18	 131,947,440	7,020,063	+2.62	Bf1 Kc1 Bg2 Kc2 Ke2 Kc1 Bh3 Kc2 Bf1 Kc1 Ke1 Kc2 Be2 Kb1 Kd1 Kxa2 Kc2 Ka1 Bf1 Ka2 Bh3 Ka1 Bg2 Ka2 Bf1 Ka1 Be2 Ka2 Bd1 Ka1 Kc1 b6 Kc2 Ka2 Be2 Ka1 Bf1 Ka2 Bg2 Ka1 Bh3 Ka2 Bf1 Ka1 Be2 Ka2 Bd1 Ka1 Kc1 Ka2
  44+	00:18	 133,210,688	7,059,433	+2.68	Bf1 Kc1 Bg2
  44+	00:19	 147,964,272	7,515,674	+2.76	Bf1 Kc1 Ke2
  44+	00:19	 149,062,123	7,544,422	+2.87	Bf1 Kb1 Kd2
  44-	00:19	 150,323,645	7,580,071	+2.80	Bf1 Kb1
  44-	00:20	 170,000,378	8,152,235	+2.59	Bf1 Kc1
  44+	00:20	 171,723,842	8,198,468	+2.73	Bf1 Kc1 Ke2
  44+	00:22	 190,812,086	8,630,926	+3.13	Bf1 Kc1 Ke2
  44	00:22	 203,672,558	8,898,278	+3.32	Bf1 Kc1 Ke2 Kc2 Bg2 Kc3 Bh3 Kc2 Bf1 d6 Ke1 Kc1 Bg2 Kc2 Ke2 Kc1 Bh3 Kc2 Bf1 Kc1 Ke1 Kc2 Be2 Kb1 Kd2 Kb2 Bd1 Kb1 Bc2+ Ka1 Kc1 Kxa2 Bd1 Ka1 Be2 Ka2 Kc2 Ka1 Bf1 Ka2 Bg2 Ka1 Bh3 Ka2 Bf1 Ka1 Be2 Ka2 Bd1 Ka1
  45-	00:24	 236,973,298	9,614,777	+3.26	Bf1 Kc1
  45+	00:26	 260,796,396	10,017,199	+3.30	Bf1 Kc1 Ke2
  45+	00:26	 261,511,663	10,026,995	+3.98	Bf1 Kc1 Ke2
  45-	00:26	 261,968,411	10,033,958	+3.80	Bf1 Kc1
  45+	00:26	 262,865,779	10,048,762	+4.00	Bf1 Kc1 Ke2
  45-	00:27	 289,852,858	10,561,008	+3.72	Bf1 Kc1
  45+	00:27	 300,498,503	10,738,968	+4.11	Bf1 Kc1 Ke2
  45+	00:28	 305,023,386	10,810,392	+5.20	Bf1 Kc1 Ke2
  45-	00:28	 309,251,649	10,877,017	+4.44	Bf1 Kc1
  45+	00:28	 317,922,575	11,014,667	+5.50	Bf1 Kc1 Ke2
  45-	00:29	 337,620,063	11,311,485	+4.02	Bf1 Kc1
  45+	00:29	 338,604,791	11,324,147	+6.09	Bf1 Kc1 Ke2
  45	00:30	 355,184,707	11,550,786	+8.68	Bf1 Kc1 Ke2 Kc2 Bg2 Kc3 Bh3 Kc2 Bf1 Kc1 Ke1 Kc2 Be2 b6 Bf1 Kc1 Ke2 Kc2 Bg2 Kc3 Bh3 Kc2 Bf1 d6 Bg2 Kc3 Bh3 Kc2 Bf1 Kc1 Ke1 Kc2 Be2 b5 Bf1 Kc1 Bh3 Kc2 Ke2 Kc3 Bg2 Kc2 Bf1 Kc1 Ke1 Kc2 Be2 d5 Bf1 Kc1 Bh3 Kc2 Ke2 Kc1 Bg2 Kc2 Bf1 Kc1 Ke1 g6 fxg6 f5
  46-	00:33	 403,319,594	12,224,897	+8.62	Bf1 Kc1
  46-	00:34	 442,358,259	12,717,919	+8.54	Bf1 Kc1
  46-	00:57	 896,676,012	15,634,935	 0.00	Bf1 Kc1
  46	00:57	 896,702,984	15,634,984	 0.00	Bf1 Kc1 Ke2 Kc2 Bg2 Kc3 Bh3 Kc2 Bf1 Kc1 Bh3 Kc2
  47+	00:57	 898,985,610	15,630,675	+0.06	Bf1 Kc1 Ke2
  47+	00:58	 899,716,318	15,629,517	+0.14	Bf1 Kc1 Ke2
  47+	00:58	 908,633,128	15,617,681	+5.25	Bf1 Kc1 Ke2
  47	01:09	1,089,775,063	15,633,074	+M82	Bf1 Kc1 Ke2 Kc2 Bh3 Kc3 Bg2 Kc2 Bf1 Kc1 Ke1 Kc2 Be2 d6 Bf1 Kc1 Ke2 Kc2 Bh3 Kc3 Bg2 Kc2 Bf1 Kc1 Ke1 Kc2 Be2 Kc1 Bd1 Kb2 Kd2 Kb1 Be2 Kb2 Bf1 Kb1 Bh3 Kb2 Bg2 Kb1 Bf1 Kb2 Be2 Kb1 Bd1 Kb2
  48	01:13	1,129,558,212	15,419,547	+M82	Bf1 Kc1 Ke2 Kc2 Bh3 Kc3 Bg2 Kc2 Bf1 Kc1 Ke1 Kc2 Be2 d6 Bf1 Kc1 Ke2 Kc2 Bh3 Kc3 Bg2 Kc2 Bf1 Kc1 Ke1 Kc2 Be2 Kc1 Bd1 Kb2 Kd2 Kb1 Be2 Kb2 Bf1 Kb1 Bh3 Kb2 Bg2 Kb1 Bf1 Kb2 Be2 Kb1 Bd1 Kb2 Bc2 Ka1
  49	02:15	1,274,671,987	9,401,450	+M82	Bf1 Kc1 Ke2 Kc2 Bh3 Kc3 Bg2 Kc2 Bf1 Kc1 Ke1 Kc2 Be2 d6 Bf1 Kc1 Ke2 Kc2 Bh3 Kc3 Bg2 Kc2 Bf1 Kc1 Ke1 Kc2 Be2 Kc1 Bd1 Kb2 Kd2 Kb1 Be2 Kb2 Bf1 Kb1 Bh3 Kb2 Bg2 Kb1 Bf1 Kb2 Be2 Kb1 Bd1 Kb2 Bc2 Ka1

Taking ideas is not a vice, it is a virtue. We have another word for this. It is called learning.
But sharing ideas is an even greater virtue. We have another word for this. It is called teaching.

User avatar
Steve Maughan
Posts: 1061
Joined: Wed Mar 08, 2006 7:28 pm
Location: Florida, USA
Contact:

Re: Mate in 82?

Post by Steve Maughan » Thu Apr 04, 2019 8:27 pm

Yes - default settings with 8 Gb of hash and six threads.
http://www.chessprogramming.net - Maverick Chess Engine

Paloma
Posts: 829
Joined: Thu Dec 25, 2008 8:07 pm

Re: Mate in 82?

Post by Paloma » Thu Apr 04, 2019 8:40 pm

Dann Corbit wrote:
Thu Apr 04, 2019 5:52 pm
I don't know how to configure ChestUCI to both find distant mates and also give me a full pv.
When I know that the mate is distant, I don't bother.

When I tell it to use the database, I get a useless response. Basically, it just says, "I found it." with no analysis.
When I tell it to use autoturbo it never seems to work on the long ones.

Perhaps I just don't know how to use it right. I also do not clearly understand the documentation for all the terms.

Now, I use Chest319 all the time (just modified it two days ago, in fact, to print out the EPD record when it is one of the mistakes it does not diagnose like "can beat the white king.") But that version would never solve checkmates in more than 12 unless you knew how to use the unusual parameters, and I don't know how to do it.

Eventually, I am going to write an AND/OR depth first proof search search using GPUs, but that is a project for when I retire, I think.
Hi Dann,

I try it with "RetroAnalysis" Backwards from Peters pgn above.
Parameters are K3/P5/X10
wich means:
- K KingSquares =3 (where the current starting square is counted too!)
- P (moving)PieceLimit =5 (black)
- X MaxMoves =10

It goes until move 12 very fast:

FEN: 8/1p1p2p1/5p1p/5P1P/1p1p1P2/pP1P1P2/P1B5/k2K4 w - - (9+9)
Positions-Analyse: C0/R0/K3/P5/X10 W:2/5
RetroAnalysis-Search for Special-Matt [C0/R0/K2/P5/X10] in 71 ... (Hash=2044MB)
Matt in 71 found ! (1 Solution in 00:00)
71/71 0:00 +M71 12.Kc1 Kxa2 13.Bd1 Ka1 14.Be2 Ka2 15.Kc2 Ka1 16.Bf1 Ka2 17.Bg2 Ka1 18.Bh3 Ka2 19.Bf1 Ka1 20.Be2 Ka2

but then M72 not found.
KingSquares at the beginning of this Study are up to five! e.g. 7th black move = 5 squares possible.
But if you increase the KingSquares these are too much variations for chest, search goes very slow.

Extract from the documentation:

Code: Select all

13) ´KingSquares´: -9..9 (Default=0)
    -------------
Also this parameter works only in the mode ´Special/Automatic´ and
means, that Chest will only examine those attacker moves, which
allow the opponent king not more than the adjusted number of squares
(where the current starting square is counted too!). ´KingSquares´
corresponds to the sometimes also ´flight squares´ or ´king mobility´
called value PLUS 1 - it is turned off by a value of ´0´ (or ´9´),
which means no restriction of the king squares, whereas the value ´1´
allows the opponent king only his current square (i.e. no king moves
are possible at all).
Negative values have a special meaning as lower limits for the search
process at ´Automatic´ mode (for detailed description see above!).

Example: Mate in 10 with ´KingSquares=2´ (Special-mode C0/R0/K2/P0/X0)
         FEN: 2K5/8/8/6Rp/8/7p/7k/7B w - -

14) ´PieceLimit´: -16..16 (Default=0)
    ------------
This parameter (also used only at ´Special/Automatic´) tells Chest
to think over such own moves only, after which the opponent is not
able to move with more pieces than the above stated value allows.
So Chest can e.g. avoid moves, which abolish blockades and pins of the
opponent, who otherwise of course would have more defense-moves again.
The value ´0´ (or ´16´) turns off this option and the value ´1´ would
leave the opponent only one movable piece.
(See also the description of the previous options!)

Example: Mate in 17 with ´PieceLimit=1´ (Special-mode C0/R0/K0/P1/X0)
         FEN: 6k1/8/5Q2/8/8/8/3p1pp1/3Kbrrb w - -

15) ´MaxMoves´: -222..222 (Default=0)
    ----------
Also this last parameter for ´Special/Automatic´ works similar to
the previous one, except that in this case Chest limits the total
number of possible opponent moves to the given value. In other words,
Chest only examines those attacker moves, after which the defender
doesn´t have more possible moves than the adjusted maximum.
A value ´0´ (or ´222´) deactivates this option completely, whereas ´1´
means the strongest restriction (to only one possible counter-move).
(See also the description of the previous options!)

Example: Mate in 17 with ´MaxMoves=2´ (Special-mode C0/R0/K0/P0/X2)
         FEN: qn6/rp1p4/pPkP1R2/K7/2P5/B1pP4/2P1P3/8 w - -

16) ´SearchOrder´: KPX / KXP / PKX / PXK / XKP / XPK / only_CR /
    -------------  only_CRK (Default=KPX)
At ´Automatic´ mode, with this option you can choose the order of
the automatic search for the optimum values of the parameters K/P/X
(the parameters C/R however will always be tested before, if they
are set to ´0´!). Of course this only refers to those parameters,
which are pre-set to ´0´ or a negative value, because with a positive
number a parameter would simply keep its adjusted value.
With the last 2 values you can adjust, that it will only be searched
for parameters C/R or C/R/K and stopped thereafter.

By a clever selection of this ´SearchOrder´ the time for automatically
finding a solution can sometimes be reduced dramatically, depending on
which of the 3 parameters K/P/X provides the maximum restrictions of
the search tree (i.e. of the possible attacker and defender moves) -
e.g. a limitation of the opponents moves or pieces will often be very
successful, which can be achieved by the setting ´XPK´. Because the
optimum search order strongly depends on the current position, it´s
not possible to automatically determine this option - so it´s really
recommended to do a short analysis of the current position, otherwise
you´ll simply have to try out, which could be the optimum setting.

Example: Mate in 20 with Automatic-mode (C/R/K/P/X = 0/0/0/0/0)
         FEN: B2k4/KPpP4/n1Pb4/P5p1/5p2/5P2/8/8 b - -
         with the default search order ´KPX´ it takes a very long time
         to find the solution, but with the reverse order ´XPK´ it´s
         found almost immediately!
Last edited by Paloma on Thu Apr 04, 2019 8:53 pm, edited 1 time in total.

Dann Corbit
Posts: 9993
Joined: Wed Mar 08, 2006 7:57 pm
Location: Redmond, WA USA
Contact:

Re: Mate in 82?

Post by Dann Corbit » Thu Apr 04, 2019 8:49 pm

Paloma wrote:
Thu Apr 04, 2019 8:40 pm
Dann Corbit wrote:
Thu Apr 04, 2019 5:52 pm
I don't know how to configure ChestUCI to both find distant mates and also give me a full pv.
When I know that the mate is distant, I don't bother.

When I tell it to use the database, I get a useless response. Basically, it just says, "I found it." with no analysis.
When I tell it to use autoturbo it never seems to work on the long ones.

Perhaps I just don't know how to use it right. I also do not clearly understand the documentation for all the terms.

Now, I use Chest319 all the time (just modified it two days ago, in fact, to print out the EPD record when it is one of the mistakes it does not diagnose like "can beat the white king.") But that version would never solve checkmates in more than 12 unless you knew how to use the unusual parameters, and I don't know how to do it.

Eventually, I am going to write an AND/OR depth first proof search search using GPUs, but that is a project for when I retire, I think.
Hi Dann,

I try it with "RetroAnalysis" Backwards from Peters pgn above.
Parameters are K3/P5/X10
wich means:
- K KingSquares =3 (where the current starting square is counted too!)
- P (moving)PieceLimit =5 (black)
- X MaxMoves =10

It goes until move 12 very fast:

FEN: 8/1p1p2p1/5p1p/5P1P/1p1p1P2/pP1P1P2/P1B5/k2K4 w - - (9+9)
Positions-Analyse: C0/R0/K3/P5/X10 W:2/5
RetroAnalysis-Search for Special-Matt [C0/R0/K2/P5/X10] in 71 ... (Hash=2044MB)
Matt in 71 found ! (1 Solution in 00:00)
71/71 0:00 +M71 12.Kc1 Kxa2 13.Bd1 Ka1 14.Be2 Ka2 15.Kc2 Ka1 16.Bf1 Ka2 17.Bg2 Ka1 18.Bh3 Ka2 19.Bf1 Ka1 20.Be2 Ka2

but then M72 not found.
KingSquares at the beginning of this Study are up to five! e.g. 7th black move = 5 squares possible.
But if you increase the KingSquares these are too much variations for chest, search goes very slow.
This is exactly the problem.
I typically analyze 100 to 10,000 positions in a batch.
I do not know what parameters to choose, and I guess it would vary from position to position.
You just have to know, I suppose, when too many king squares are in play and things of that nature.

Is there a way to have ChestUCI figure things out for itself and analyze?
That's what I need and it is why I do not use it for deep problems.
I do not know the parameters to choose.
I do not want to set up 1000 different sets of parameters.
Taking ideas is not a vice, it is a virtue. We have another word for this. It is called learning.
But sharing ideas is an even greater virtue. We have another word for this. It is called teaching.

JohnS
Posts: 134
Joined: Sun Feb 24, 2008 1:08 am

Re: Mate in 82?

Post by JohnS » Thu Apr 04, 2019 9:20 pm

Dann Corbit wrote:
Thu Apr 04, 2019 5:52 pm


Eventually, I am going to write an AND/OR depth first proof search search using GPUs, but that is a project for when I retire, I think.
We need a new solver for long mates. Would next week be too soon for you to retire 😀

Dann Corbit
Posts: 9993
Joined: Wed Mar 08, 2006 7:57 pm
Location: Redmond, WA USA
Contact:

Re: Mate in 82?

Post by Dann Corbit » Thu Apr 04, 2019 9:23 pm

JohnS wrote:
Thu Apr 04, 2019 9:20 pm
Dann Corbit wrote:
Thu Apr 04, 2019 5:52 pm


Eventually, I am going to write an AND/OR depth first proof search search using GPUs, but that is a project for when I retire, I think.
We need a new solver for long mates. Would next week be too soon for you to retire 😀
Fraid so. I need to pay off all the debts from my wife's medical bills.
I guess 5 years, but it is hard to say.
Taking ideas is not a vice, it is a virtue. We have another word for this. It is called learning.
But sharing ideas is an even greater virtue. We have another word for this. It is called teaching.

JohnS
Posts: 134
Joined: Sun Feb 24, 2008 1:08 am

Re: Mate in 82?

Post by JohnS » Thu Apr 04, 2019 9:27 pm

Dann Corbit wrote:
Thu Apr 04, 2019 9:23 pm
JohnS wrote:
Thu Apr 04, 2019 9:20 pm
Dann Corbit wrote:
Thu Apr 04, 2019 5:52 pm


Eventually, I am going to write an AND/OR depth first proof search search using GPUs, but that is a project for when I retire, I think.
We need a new solver for long mates. Would next week be too soon for you to retire 😀
Fraid so. I need to pay off all the debts from my wife's medical bills.
I guess 5 years, but it is hard to say.
Sorry to hear about your wife and I hope things are ok with her now.

peter
Posts: 1778
Joined: Sat Feb 16, 2008 6:38 am
Full name: Peter Martan

Re: Mate in 82?

Post by peter » Fri Apr 05, 2019 5:44 am

Dann Corbit wrote:
Thu Apr 04, 2019 5:52 pm
Eventually, I am going to write an AND/OR depth first proof search search using GPUs, but that is a project for when I retire, I think.
Olaf Jenkner's Gustav (3.3, not the latest one anymore, as far as I know) gives solution in 51 seconds with automtic parameters, only given 82 moves to search for.

1.Le2-f1 Kc2-c1 2.Ke1-e2 Kc1-c2 3.Lf1-g2 Kc2-c1 4.Lg2-h3 Kc1-c2 5.Lh3-f1 Kc2-c1 6.Ke2-e1 Kc1-b2 7.Ke1-d1 Kb2-b1 8.Kd1-d2 Kb1-b2 9.Lf1-e2 Kb2-b1 10.Le2-d1 Kb1-b2 11.Ld1-c2 Kb2-a1 12.Kd2-c1 Ka1-a2 13.Lc2-d1 Ka2-a1 14.Kc1-c2 Ka1-a2 15.Ld1-e2 Ka2-a1 16.Le2-f1 Ka1-a2 17.Lf1-g2 Ka2-a1 18.Lg2-h3 Ka1-a2 19.Lh3-f1 Ka2-a1 20.Lf1-e2 Ka1-a2 21.Le2-d1 Ka2-a1 22.Kc2-c1 Ka1-a2 23.Ld1-c2 Ka2-a1 24.Lc2-b1 Bb7-b6 25.Lb1-c2 Ka1-a2 26.Lc2-d1 Ka2-a1 27.Kc1-c2 Ka1-a2 28.Ld1-e2 Ka2-a1 29.Le2-f1 Ka1-a2 30.Lf1-g2 Ka2-a1 31.Lg2-h3 Ka1-a2 32.Lh3-f1 Ka2-a1 33.Lf1-e2 Ka1-a2 34.Le2-d1 Ka2-a1 35.Kc2-c1 Ka1-a2 36.Ld1-c2 Ka2-a1 37.Lc2-b1 Bb6-b5 38.Lb1-c2 Ka1-a2 39.Lc2-d1 Ka2-a1 40.Kc1-c2 Ka1-a2 41.Ld1-e2 Ka2-a1 42.Le2-f1 Ka1-a2 43.Lf1-g2 Ka2-a1 44.Lg2-h3 Ka1-a2 45.Lh3-f1 Ka2-a1 46.Lf1-e2 Ka1-a2 47.Le2-d1 Ka2-a1 48.Kc2-c1 Ka1-a2 49.Ld1-c2 Ka2-a1 50.Lc2-b1 Bd7-d6 51.Lb1-c2 Ka1-a2 52.Lc2-d1 Ka2-a1 53.Kc1-c2 Ka1-a2 54.Ld1-e2 Ka2-a1 55.Le2-f1 Ka1-a2 56.Lf1-g2 Ka2-a1 57.Lg2-h3 Ka1-a2 58.Lh3-f1 Ka2-a1 59.Lf1-e2 Ka1-a2 60.Le2-d1 Ka2-a1 61.Kc2-c1 Ka1-a2 62.Ld1-c2 Ka2-a1 63.Lc2-b1 Bd6-d5 64.Lb1-c2 Ka1-a2 65.Lc2-d1 Ka2-a1 66.Kc1-c2 Ka1-a2 67.Ld1-e2 Ka2-a1 68.Le2-f1 Ka1-a2 69.Lf1-g2 Ka2-a1 70.Lg2-h3 Ka1-a2 71.Lh3-f1 Ka2-a1 72.Lf1-e2 Ka1-a2 73.Le2-d1 Ka2-a1 74.Kc2-c1 Ka1-a2 75.Ld1-c2 Ka2-a1 76.Lc2-b1 Ba3-a2 77.Lb1-c2 Bg7-g5 78.Bh5-g6 Bh6-h5 79.Bg6-g7 Bh5-h4 80.g7-g8D Bh4-h3 81.Dg8-g1 Bh3-h2 82.Dg1-d4#
Peter.

User avatar
Steve Maughan
Posts: 1061
Joined: Wed Mar 08, 2006 7:28 pm
Location: Florida, USA
Contact:

Re: Mate in 82?

Post by Steve Maughan » Fri Apr 05, 2019 11:42 am

It's interesting that Komodo seems to excel in this position. Clearly Mark and Larry have devised some clever logic. The position reminds me of the game Jonny vs. Komodo from the WCCC in 2015; a blocked game where Komodo was eventually out maneuvered. I wonder if Mark and Larry developed the heuristics that work so well in this position after losing the game to Jonny?

http://www.chessprogramming.net - Maverick Chess Engine

Post Reply