Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Approximate # of Unique chess positions (FEN working)

Author: Sune Fischer

Date: 02:21:43 01/19/02

Go up one level in this thread


On January 19, 2002 at 03:02:41, Uri Blass wrote:
>>>>>>>Does this take into account all the possible promotions? It is possible to have
>>>>>>>9 queens on one side or 9 rooks or 9 bishops or 9 knights, or any combination
>>>>>>>thereof. Each possible promotion will make the number larger. 31 pieces on the
>>>>>>>board has more possible positions than 32 because you have to take into account
>>>>>>>the promotions. If 31 pieces are remaining there is upto 3 possible promotions.
>>>>>>>Makes the calculation a little more difficult.
>>>>>>>
>>>>>>>Chris
>>>>>>>
>>>>>>
>>>>>>Oops, make that 9 queens or 10 rooks, or 10 bishops or 10 knights. Some days my
>>>>>>fingers work faster than my brain.
>>>>>>
>>>>>>Chris
>>>>>
>>>>>You do have a point Chris.
>>>>>I tried to refine the number a little bit:
>>>>>
>>>>>All 32 pieces:
>>>>>(64*(64 - 4)*23^4*21^4*18^4*14^4*62!)/((62 - 14)!*(8!*2!*2!*2*2)^2)
>>>>>= 5.13*10^36
>>>>>
>>>>>I put in the factor: 23^4*21^4*18^4*14^4
>>>>>because the d and e pawns has 23 squares, the c and f pawns has 21 squares and
>>>>>the b and g has 18, a and h has 14 (unless I miscounted;)
>>>>
>>>>I do not understand it
>>>
>>>Actually there are 24 squares for a white d pawn:
>>8/pppppppp/PPPPPPPP/PPPPPPP1/1PPPPP2/2PPP3/3P4/8 w - - 0 1
>
>I understand now your idea but with 32 pieces there are only 5 options for white
>d pawns because the white d pawn can go to another file only by capture.
>
>d2 d3 d4 d5 d6 d7 are the options.
>
>My 15 options are for white and black pawn on the d file and the same for every
>file.
>
>d2 d7
>d2 d6
>d2 d5
>d2 d4
>d2 d3
>d3 d7
>d3 d6
>d3 d5
>d3 d4
>d4 d5
>d4 d6
>d4 d7
>d5 d6
>d5 d7
>d6 d7
>
>are the possible squares for d pawns when there are no captures.
>
>It is exactly 15 and the same for every file.
>
>My program did not use the 15^8 limit for the pawns squares and assumed that
>every pawn can be chosen from 48 squares.
>
>It is not close to be truth when you have 32 pieces but it becomes closer to the
>truth when there are less pieces.
>
>It is clear that if both sides have exactly 2 pawns you can choose them from the
>48 squares in the board in every possible way.
>
>It is not truth for 3 pawns and h2 h3 g2 is an example for squares that cannot
>be squares of white pawns but it is still truth for almost all the cases of 3
>pawns.
>
>Uri

Oh, so that's what you mean, well I'm not concerned with counting *too many*
squares since is was supposed to be an upper bound anyway. That it was easier
when the capturings began.

I did think of one interesting example;
24 pieces with 12 pawn promotions (I think that is possible):
2*4^12*(64*(64 - 4)*62!)/((62 - 22)!*(2!*3!*3!*2*2)^2)
=5.99*10^43

It is the highest number I can produce;)

Depending on what the question is, we can reduce a bit by using the 8
symmetries, or increase by side-to-move and en-passent rights etc.
Althought the latter is what is needed for hashing, then when I think about it
does seems a bit silly to count a position like this 192 times:

[D]r3k2r/8/8/ppp5/5PPP/8/8/R3K2R w KQkq - 0 1

-S.



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.