Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Rep detection answer

Author: Tim Foden

Date: 23:53:59 04/22/02

Go up one level in this thread


On April 22, 2002 at 15:22:39, Will Singleton wrote:

>On April 22, 2002 at 02:29:40, Tim Foden wrote:
>
>>On April 21, 2002 at 19:29:39, Will Singleton wrote:
>>
>>>All,
>>>
>>>[D]5nk1/2R3p1/r4p2/4P3/p2R2N1/7P/1r4P1/7K b - - 0 42
>>>
>>>The question is, my prog wants to play a3, and doesn't see the forced repetition
>>>until ply 9.  I've just thrown away my old buggy rep code, and replaced it with
>>>a dedicated hash table.  The new stuff seems to be debugged and working (maybe),
>>>but there must be a trick to getting this position earlier.  An extension,
>>>perhaps?
>>
>>Green Light sees this at 5 ply.  The line it gives is:
>>42... a3 43. Nxf6+ gxf6 44. Rg4+ Kh8 45. Rh4+ Kg8 46. Rg4+
>>
>>The only extension used here is a check extension x 4, which gets it to the
>>required 9 ply.  The other trick, of course, is that GL considers it a draw if
>>the position occurs twice in the current search.
>>
>
>Ok, found a bug having to do with null move.  Always nice to kill bugs.

It sure is.  It's sometimes alarming to find them and realise how long they have
been there. :)  I'm fairly sure there's one in GL that I've been trying to track
down for months...  just try Fine 70.  :)

>  2      90     0.02       339    10140  Ne6 Rc8 Kf7
>  3      63     0.02       913    27360  Ne6 Rc8 Kh7 Rd7
>  3      69     0.02      1183    35460  Rb1 Kh2 Ne6 Rc8 Kf7
>  4      42     0.04      3208    48120  Rb1 Kh2 Ne6 Rc8 Kh7 Rd7
>  5      14     0.12     18176    90840  Rb1 Kh2 Pxe5 Nxe5 a3 Rd8
>  5     102     0.24     40963   102360  a3 Rd1 a2 Rc1 Nh7 Pxf6 Nxf6
>  6       0     0.49     91554   112080  a3 Nxf6 Pxf6 Rg4 Kh8 Rh4 Kg8 Rg4
>  6      67     1.06    126419   114900  Rb1 Kh2 a3 Pxf6 a2 Rxg7 Kh8 Rd8 Rxf6
>  7      54     1.44    203988   117660  Rb1 Kh2 Pxe5 Nxe5 a3 Rg4 Ne6 Rd7
>  7      90     2.25    289403   119700  Pxe5 Nxe5 a3 Rg4 g6
>  8     111     3.30    432827   123660  Pxe5 Nxe5 a3 Rd1 a2 Rc1 Re2 Nc4 Rf6
>
>
>btw, gl chess looks real strong now.

Thanks.  I'm currently trying quite hard to make GL stronger.  But as you know,
many of the things you try turnout to have the opposite effect.

>I'd like to see that algorithm info you

The algorithm info describes some the things I needed to know when I was first
developing GL.  e.g.  It has some hints on hash tables (you need to store items
for both black-to-move and white-to-move, etc.)  It also describes in a general
way how an old version of GL works.  I put it on a web page so that others may
be able to benefit from it.

It is very out of date now, and I don't think it says anything new.

>mention on your web site, perhaps it's in the download?

The download includes a file called changes.txt, which hash a list of the
changes made in each realeased version.

Cheers, Tim.



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.