Author: leonid
Date: 15:54:35 07/07/01
Go up one level in this thread
On July 07, 2001 at 18:37:06, leonid wrote:
>On July 07, 2001 at 16:27:46, Heiner Marxen wrote:
>
>>On July 07, 2001 at 12:47:12, leonid wrote:
>>
>>>Hi!
>>>
>>>If you would like to solve mate then position then look into this:
>>>
>>>[D]q2k2q1/1r1q1r2/nRqQqRn1/NqQbQqN1/QQ1K1QQ1/Q2B2Q1/p2b2p1/8 w - -
>>
>>Hi Leonid,
>
>Hi!
>>
>>According to Chest this is a mate in ten with one solution move:
>>
>>PV: Qxd7+ Qcxd7 Qb8+ Qxb8 Qxb8+ Rxb8 Rxb8+ Nxb8 Qxb8+ Qc8 Nxe6+ Bxe6 Qcb6+ Rc7
>>Q6xc7+ Ke8 Qxg6+ Qf7 Qce7#
>>
>>(K7/600, 350 MB hash, 2.3 hours)
>>
>>>This position is curious from the point of view that its branching factor (when
>>>mate is searched by brute force) diminish with the depth of search. Usually it
>>>is contrary that you can see.
>>
>> depth time EBF
>># 2 0.01s 0kN 1.00 1- 0
>># 3 0.04s [ 4.00] 1kN 0.97 87- 0
>># 4 0.37s [ 9.25] 17kN 1.03 1272- 0
>># 5 2.62s [ 7.08] 126kN 1.16 11441- 0
>># 6 16.15s [ 6.16] 800kN 1.35 77134- 0
>># 7 90.26s [ 5.59] 4638kN 1.59 435750- 0
>># 8 524.25s [ 5.81] 28147kN 1.88 2418963- 217
>># 9 2157.76s [ 4.12] 116264kN 2.44 9713815- 1536836
>># 10 8206.47s [ 3.80] 441956kN 3.20 36849208- 28101307
>>
>>Yes, for Chest also the effective branching factor (EBF) is decreasing.
>
>So this tendency really depend on position. Mine went like this when search is
>done by brute force:
>
4 moves - 1.1 sec 52k (NPS)
6.65 (branching factor)
5 moves - 7.3 sec 42k
6.02
6 moves - 44 sec 42k
5.91
7 moves - 4 min 20 sec 42k
5.18
8 moves - 22 min 31 sec 50k
4.4
9 moves - 1h 39m 6sec 54k
10 moves - 2h 37m 14sec
>>> Even second variation of this position, that I
>>>tried by brute force, have the same tendency. Second position was left as more
>>>tough one. If you want to see it then change position of two pieces on the
>>>board. White bishop from D3 goes to H7 and black bishop from D2 to H2.
>>
>>I have not yet tried that one.
>>
>>> I wonder
>>>even if the second position don't have some solution for blacks but have no time
>>>to try it. If you will try for blacks, please say me your result if even your
>>>result was negative.
>>
>>I'm not sure what you are talking about. Same board, but with black to move
>>and mate?
Yes!
I expect to put this night this position with blacks. It can take a lot of time.
I could see that Angrim found response for whites by selective in much shorter
time that mine. My program went for 7 hours in selective and found it in 13
moves. This is why I expect that search for blacks could be also long.
If you will start searching for white side, you can start directly from 10
moves. First 9 moves I looked already by brute force. No mate in sight.
Cheers,
Leonid.
>>>Please say your result.
>>>
>>>Thanks,
>>>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.