Author: Dieter Buerssner
Date: 08:32:18 01/31/02
Go up one level in this thread
On January 30, 2002 at 23:28:50, Robert Hyatt wrote:
>On January 30, 2002 at 11:52:45, Albert Silver wrote:
>
>>I don't understand. If the search at that point doesn't get a tablebase hit with
>>a KQQQ vs. K and instead gets a score of +30 pawns, it will somehow contrive to
>>play worse?? Or make a poorer choice? Say opt for an alternate line that gives
>>it +35 pawns instead of the direct mate?
>>
>
>First, that screws up the alpha/beta search. If _all_ the scores are near a
>mate score, and then a few drop way down, that widens the search window and
>makes the search go slower. If you are already winning, fine. But if you
>are not, this can hurt.
I think, you are either in all cases already winning if you would hit some KQQXK
tables, or the result from this table does not really matter. But as you note,
it will help to avoid searching some nodes. I think it is not totally clear, if
the additional TB hits need less time, than the search avoided. You will most
probably get in the second next ply a cutoff by a four man table anyway. Even if
not, and in the winning case, your alpha will still show some massive advantage,
and I see no danger how this could hurt in a normal play (if you already have an
alpha of +30, nothing can happen anymore, even when you search few nodes more).
At least, when the engine has no serious bug, for example missing a stalemate.
>Also you can look silly... would you _really_ want to hit a KNN vs KP ending
>and possibly draw, because the tables say MATE in N, rather than hitting a
>KQQQ vs K which says +30 rather than mate in 9 or whatever?
It seems almost impossible, that you can force a long comlicated mate in one
line, and force KQQQK in another line from the same root position. BTW. The
longest mate of KQQQK is mate in 3, so for most chess engines in the worst case
a 5 ply search schould be enough.
I have almost not tested 4+1 TBs, because I only have one installed to test
basic funtionality. IIRC, in few tests I got sent by email, the performance was
never better than without 4+1 TBs.
Interesting is also Eugene Nalimov's comment in probe.txt:
- T41_INCLUDE - if you want to probe any 5-man 4+1 table (e.g. KPPPK);
if you will not ever probe those tables (I suppose
most chess program will never do that), you can save
some space by not including corresponding code and
enumeration tables;
If I understand this correctly, Eugene also doubts the usefulness of the 4+1
tables. I may add, that also the EGTB-cache may be used more efficiently without
probing 4+1 tables.
>>> Yes a
>>>program should win KQQQK or KRRRK or KBBBK or KNNNK, but it helps to know
>>>_exactly_ how deep the mate is, to make the search as efficient as possible.
I think, in practical games this should never be relevant for the result. It may
help to get a mate score faster instead of a big advantage. Similar things I see
with my limited set of TBs. Sometimes the line ends in a TB position that I do
not have. So sometimes I see scores of +20, where I could see a mate score.
Regards,
Dieter
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.