Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Egtb mates and upper bounds

Author: Hermano Ecuadoriano

Date: 08:28:28 02/05/01

Go up one level in this thread


On February 05, 2001 at 10:57:14, Hermano Ecuadoriano wrote:

>On February 05, 2001 at 09:46:59, Paul wrote:
>
>>Hi!
>>
>>One last message from me on the subject, since I think I know now what made you
>>hesitant and wanted to share this to help clear out all the confusion forever.
>>
>>1. When a program finds a mate in *favor* of itself, so mating the opponent,
>>it's immediately an upper bound on the real matescore, so with more time the
>>announcement stays the same or gets lower (M78 ... M54 ... etc).
>>
>>2. When a program finds a mate *against* itself, so mating itself, and it stays
>>that way until the end of ply (when all the moves of that ply have been
>>examined), it's also an upper bound on the real matescore, so with more time the
>>announcement stays the same or gets lower (M78 ... M54 ... etc).
>>
>>Now, there are 2 interesting cases:
>>
>>1. Before the winning side finds a mate in favor of itself, it can even find a
>>mate against itself, but that will have disappeared by the end of the ply it
>>found this mate on.
>>
>>2. Before the losing side finds a mate against itself, it can have found lower
>>mates against itself, but again, those will have disappeared by the end of the
>>ply in question.
>>
>>Hope this helps a bit!
>>
>>Groetjes,
>>Paul
>
>I gave a clear counterexample.

O.K. I understand the question raised about the crafty output I snipped.
To save my credibility, I had to figure out why that -> wasn't there.

With noise= 1, to show everything, the .log shows

snip
Black(1): go
end-game phase
              clearing hash tables
              time surplus   0.00  time limit 10:00 (10:00)
         nss  depth   time  score   variation (1)
                1->   0.01   0.00   Ng5
                2     0.02   0.00   Ng5 Bb8
                2->   0.03   0.00   Ng5 Bb8
                3     0.03   0.00   Ng5 Bb8 Kh7
                3->   0.06   0.00   Ng5 Bb8 Kh7
                4     0.13   0.00   Ng5 Bb8 Kh7 Bc7
                4->   0.14   0.00   Ng5 Bb8 Kh7 Bc7
                5     0.28   0.01   Ng5 Bf4 Nh3 Bxh6+ <HT>
                5     0.31   0.00   Kg8 Be5 Ng5 Be8 Nh3
                5->   0.31   0.00   Kg8 Be5 Ng5 Be8 Nh3
                6     0.33   0.00   Kg8 Be5 Ng5 Be8 Nh3 Bd7
                6->   0.35   0.00   Kg8 Be5 Ng5 Be8 Nh3 Bd7
                7     0.68   0.01   Kg8 Be5 Ng5 Bf4 Kg7 Bxg5 <HT>
                7->   0.78   0.01   Kg8 Be5 Ng5 Bf4 Kg7 Bxg5 <HT>
                8     0.92   0.01   Kg8 Be5 Ng5 Bf4 Kg7 Bxg5 <HT>
                8->   0.93   0.01   Kg8 Be5 Ng5 Bf4 Kg7 Bxg5 <HT>
                9     1.21     --   Kg8
                9     1.54  Mat68   Kg8 Kg6 Ng5 Be2 Ne6 Kxh6 <HT>
                9     2.06     ++   Ng5!!
                9     2.31   0.01   Ng5 Be5+ Kh7 Bf4 Nh3 Bxh6 <HT>
                9->   2.34   0.01   Ng5 Be5+ Kh7 Bf4 Nh3 Bxh6 <HT>
               10     3.34   0.01   Ng5 Be5+ Kh7 Bf4 Nh3 Bxh6 <HT>
               10->   3.39   0.01   Ng5 Be5+ Kh7 Bf4 Nh3 Bxh6 <HT>
snip

With noise= 100000, which I did only to simplify the example, the .log shows:

snip
Black(1): go
end-game phase
              time surplus   0.00  time limit 1000:00 (1000:00)
         nss  depth   time  score   variation (1)
                9     0.70  Mat68   Kg8 Kg6 Ng5 Be2 Ne6 Kxh6 <HT>
               10     1.34   0.01   Ng5 Be5+ Kh7 Bf4 Nh3 Bxh6 <HT>
               10->   1.35   0.01   Ng5 Be5+ Kh7 Bf4 Nh3 Bxh6 <HT>
snip

I posted this in total innocence.
I don't know why the -> line isn't there.
I was trying to be helpful. I found what I thought was a counterexample,
and tried to explain it.
I mean it this time: I quit posting anything except my opinions.
Best wishes.




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.