Author: leonid
Date: 05:41:45 02/03/02
Go up one level in this thread
On February 03, 2002 at 06:19:35, Heiner Marxen wrote:
>On February 02, 2002 at 07:37:27, leonid wrote:
>
>>[D]2QqQN2/1q3q1R/B1r1RPq1/n1k2N1q/Rq1rq1bP/1Q2P1B1/2q2b1q/K2nB3 w - -
>>
>>Please indicate your result.
>>
>>Thanks,
>>Leonid.
>
>Hello Leonid,
>
>This one is surprisingly accessible for Chest: on my Athlon/600 with 350 MB
>hash it needs only 65.8 minutes to find that this is a mate in 12 with
>two key moves:
>
>PV: Bxb4+ Qxb4 Re5+ Qed5 Nd7+ Qdxd7 Qf8+ Qxf8 Qxf8+ Rd6 Rxd5+ Kc6 Bb7+ Kxb7
>Rxd7+ Qc7 Rxb4+ Rxb4 Qxb4+ Rb6 Rxc7+ Ka6 Rxa5#
>
>PV: Re5+ Qed5 Nd7+ Qdxd7 Bxb4+ Qxb4 Qf8+ Qxf8 Qxf8+ Rd6 Rxd5+ Kc6 Bb7+ Kxb7
>Rxd7+ Qc7 Rxb4+ Rxb4 Qxb4+ Rb6 Rxc7+ Ka6 Rxa5#
>
>After the first 3 moves both transpose to the same position/solution.
>The timing shows a very good EBF: it is shrinking all the time down to
>below 3:
>
># 3 0.01s 1kN [ 14.57] 0.96 62- 0
># 4 0.11s [ 11.00] 7kN [ 8.14] 1.10 483- 0
># 5 0.46s [ 4.18] 35kN [ 5.23] 1.43 2129- 0
># 6 2.41s [ 5.24] 159kN [ 4.58] 2.07 11344- 0
># 7 9.97s [ 4.14] 618kN [ 3.89] 3.39 49756- 0
># 8 38.48s [ 3.86] 2275kN [ 3.68] 5.23 203391- 0
># 9 113.41s [ 2.95] 6544kN [ 2.88] 8.08 632407- 0
># 10 421.92s [ 3.72] 24007kN [ 3.67] 9.20 2304940- 158
># 11 1426.10s [ 3.38] 77337kN [ 3.22] 10.89 7953425- 564639
># 12 3946.84s [ 2.77] 202561kN [ 2.62] 15.53 24840756- 16092886
>
>Also, the hash table efficiency appears to be better than normally: speed up
>factor 15.53 for depth=12. This could be the main reason for the small EBF.
>Finally, in 55.9% of the cases the defender does defend by mating the attacker
>in just 1 move, cutting off the search tree.
Hello, Heiner!
As usual, this kind of mate position is easy only for your solver and for nobody
else. I can only expect that some other program will come with selective search
solution but almost zero hope for brute force. This kind of position, just
beside your mate solver, is out of reach of everybody.
With mine Celeron 600Mhz and with no hash, I even not went to solve all 12 moves
by brute force. If not, it could take one next night to make a search. Mine
solved it by selective in 12, and by brute force only 11 moves. This way I was
sure in advance that this position is mate in 12.
Brute force data:
Move Time Branching factor NPS
4 0.49 sec 162k
6.11
5 3.02 sec 109k
5.98
6 18.07 sec 79k
5.12
7 1 min 32 sec 70k
4.24
8 6 min 33 sec 60k
3.5
9 23 min 00 sec 59k
4.5
10 1 h 43 min 29 sec 67k
3.8
11 6 h 33 min 20 sec 74 k
Cheers,
Leonid.
>Cheers,
>Heiner
This page took 0 seconds to execute
Last modified: Thu, 15 Apr 21 08:11:13 -0700
Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.