Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Try to solve mate.

Author: leonid

Date: 16:39:32 09/14/02

Go up one level in this thread


On September 14, 2002 at 16:58:23, Heiner Marxen wrote:

>On September 14, 2002 at 10:44:49, leonid wrote:
>
>>On September 14, 2002 at 10:14:52, Paul wrote:
>>
>>>On September 14, 2002 at 09:47:21, leonid wrote:
>>>
>>>>On September 14, 2002 at 09:26:25, leonid wrote:
>>>>
>>>>>On September 14, 2002 at 08:51:31, Paul wrote:
>>>>>
>>>>>>On September 14, 2002 at 07:55:35, leonid wrote:
>>>>>>
>>>>>>>Hi,
>>>>>>>
>>>>>>>This mate is easy and only 9 moves deep:
>>>>>>>
>>>>>>>[D]bqRNNkqr/Q7/qPnPqBqN/Q4q2/5Qr1/5q2/5nQ1/bQqQqBKR w - -
>>>>>>
>>>>>>Hi Leonid,
>>>>>>
>>>>>>Pretz (p3/1000) also finds this one easy, it gets this in 3 seconds:
>>>>>>
>>>>>>00:03 WM9 06 Nxe6+ Qgxe6 Be7+ Nxe7 dxe7+ Qxe7 Qaxf5+ Bf6 Nxf6+ Qaxc8 Nh7+ Ke8
>>>>>>Qxg6+ Rxg6 Qgxg6+ Qf7 Qaxf7#
>
>Chest (Athlon 1500+, 128MB hash) confirms this in 56.21 secs:

Hi, Heiner!

>PV: Nxe6+ Qgxe6 Bg7+ Bxg7 Qbxf5+ Bf6 Nxf6+ Qaxc8 Nh7+ Ke8 d7+ Kd8 dxc8=Q+ Kxc8
>Qfc7+ Qxc7 Qxc7#
>
> Nxe6+ Qgxe6 Bg7+ Bxg7 Qbxf5+ ...+6
>                  Qxg7 Qbxf5+ ...+3
>       Qexe6 Be7+ =*=  dxe7+  ...+5
>       Qfxe6 Bg7#

Probably our branching were close even if my initial speed (before 4 moves)
should be better for You.

I looked brute force only 8 moves in order to know that it is really 9 moves
position. It, on my usual setting (Celeron 600Mhz, no hash) was:

Move            Time            Branching factor        NPS

4               0.27 sec                                125k
                                4.6
5               1.26 sec                                105k
                                5.34
6               6.75 sec                                80k
                                4.95
7               33.46                                   79k
                                5.08
8               2 min 50 sec



>It is so easy/fast, because black has mating resources, too, like Rxg2#.
>
>>>>>>>Still, maybe, You will be tempted for something more heavy and I this way will
>>>>>>>be able to find more on this position:
>>>>>>>
>>>>>>>[D]2qRQq2/p6P/q1BqqR1N/r6B/n1QNQQ1B/r6k/b1QKNP1q/bq4qn w - -
>>>>>>>
>>>>>>>On description of this mate I can see that it was solved by selective in 13
>>>>>>>moves but brute force search went not up to the end in order to find shortest
>>>>>>>move. Brute force search, without finding any mate, went only 9 moves deep.
>>>>>>
>>>>>>This one was indeed a lot more difficult, mine didn't fare so well here:
>>>>>>
>>>>>>06:49 WM11 07 Q8xe6+ Qdxe6 Qff3+ Rxf3 Qxf3+ Kxh4 Nhf5+ Rxf5 Nxf5+ ...
>>>>>>
>>>>>>So it found a mate in 11 in almost 7 minutes, with an incomplete pv ... these
>>>>>>d#*&#@rned hashtables ;) ... will have to think about this ... any tips anyone?
>>>>>>
>>>>>>>Please indicate Your result.
>>>>>>>
>>>>>>>Thanks,
>>>>>>>Leonid.
>>>>>>
>>>>>>Thanks for your mates as usual ...
>>>>>
>>>>>Thanks, Paul!
>>>>>
>>>>>Now I can see that second one is no far then 11 moves. I  can even put this
>>>>>position at 10 moves by brute force for night. If braching factor will not
>>>>>suddenly jump, then it should be enough time for me for finding its depth.
>>>>>
>>>>>First position have excellent time!
>>>>>
>>>>>Mine, on usual setting, found mate in 1.75 sec.
>>>>>
>>>>>Cheers,
>>>>>Leonid.
>>>>
>>>>Paul, just after writing my message, I started to see second position in my file
>>>>to indicate 11 moves. When I finally found it (I not indicate before on what
>>>>page I had it), I looked that it was in reality solved by selected in 11 moves,
>>>>just as You said. Before upper part of page was covered with page keeper and it
>>>>is how I missed some additional description. If I knew this yesterday, when I
>>>>found this position, I could put search for 10 moves by brute force at night.
>>>
>>>Hey!! ... so you tricked us ... that's not fair ... I would have found it much
>>>faster if I'd known! ... just kiddin' ... :)
>>
>>I was actually tricked by myself. I did expected that this position was very
>>difficult since I could see that it was solved by very slow selective and only
>>at 13 moves. 13 moves at large default setting (I use it before searching by
>>much easier solution) took 13.5 minutes to solve. When I looked into 11 moves
>>selective solution its paramether indicated that solution was almost simple.
>>
>>
>>>>Selective in 11 moves was solved in 18 seconds. Usual setting.
>>>
>>>Not bad at all ... you really should implement hash to get a bit faster though
>>>... ;)
>
>Chest is still running for depth 11, but after 3478 secs has already reported
>that there is no mate in 10.  So, we are already sure, that 11 is minimal
>depth.

Paul just said to me that there are no mate in 10.

>#  4      0.09s                17kN [ 10.28]  1.07        737-         0
>#  5      0.96s [ 10.67]      137kN [  7.88]  1.21       8654-         0
>#  6      6.32s [  6.58]      791kN [  5.79]  1.49      66512-         0
>#  7     31.30s [  4.95]     3675kN [  4.65]  2.11     351756-         0
>#  8    143.98s [  4.60]    16349kN [  4.45]  2.73    1718242-      7352
>#  9    668.36s [  4.64]    73337kN [  4.49]  3.46    8452143-   5252957
># 10   3477.74s [  5.20]   373030kN [  5.09]  4.02   45004182-  41804949


On mine, with usual setting:

Move              Time            Branching factor         NPS

4                 1.26 sec                                 113k
                                  8.0
5                 10.16 sec                                67k
                                  7.52
6                 1 min 16 sec                             67k
                                  6.28
7                 7 min 58                                 57k
                                  5.21
8                 41 min 35 sec                            59k
                                  5.21
9                 3h 37m 14s                               60k

Cheers,
Leonid.


>Cheers,
>Heiner
>
>
>>I do expect that one day I will come back to my programming. At least, this last
>>week I had plaintly of work and it could be that needed second computer will be
>>not that long to wait for. Still 64 bits gadget will be ideal to start with. I
>>am sure that when new, 64 bits, computer will arrive chatting of people here
>>will be very animated. Too many things to redo, or write from scratch.
>>
>>Cheers,
>>Leonid.
>>
>>>>Cheers,
>>>>Leonid.
>>>
>>>Groetjes,
>>>Paul



This page took 0.09 seconds to execute

Last modified: Thu, 07 Jul 11 08:48:38 -0700

Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.