Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Some thoughts on Dann Corbit's rotated alternative

Author: Dann Corbit

Date: 14:45:10 02/27/06

Go up one level in this thread


On February 27, 2006 at 15:13:59, Gerd Isenberg wrote:

>On February 27, 2006 at 14:16:38, Dann Corbit wrote:
>
>>You have a special genius for bitboards and I always enjoy reading your posts.
>
>Thanks Dann,
>
>i appreciate that - but i have not special genius but morbus knuth ;-)
>I like interactive brainstorming and to formulate ideas so that they become more
>clear. I had some difficulties to understand your switch approach first.
>
>What do you think about masking off outer occupied bits?
>As well inside your switch cases as well as a 64-bit factor of a De Bruijn
>multiplication?
>
>They are redundant for attacks and also for possible xrays.
>It reduces the number of occupied cases to max 2**(raylenght-2) if source square
>is an outer square - or most often to 2**(raylenght-3), if source is some inner
>square, for instance bishop on e5 on the a1-h8 diagonal:
>
>0 0 0 0 0 0 0 x
>0 0 0 0 0 0 1 0
>0 0 0 0 0 1 0 0
>0 0 0 0 b 0 0 0
>0 0 0 1 0 0 0 0
>0 0 1 0 0 0 0 0
>0 1 0 0 0 0 0 0
>x 0 0 0 0 0 0 0
>
>Thus some 64-states and most 32-bit or less states fits perfectly to a 2**6
>range by "De Bruijn folding" with empirical determined individual magics for
>each source square and direction.
>
>Obviously each single bit-subset of the five or six bits leaves unique upper six
>bits after a multiplication with a DeBruijn constant, because the DeBruijn is
>per definition a sequence of 64 unique 6-bit strings - and mutiplication with
>power of two is like shifting left by log2. That is the original idea of using
>De Bruijn multiplication as log2 or bitscan.
>
>Having more bits set produces modulo 64 sums of appropriate unique subsequences.
>So we have "only" to choose a De Bruijn or "modified" De Bruijn where all sums
>modulo 64 are unique.

I am not really sure that I understand your suggestion about ignoring outer
occupied bits.

I need to x-ray for some depth for battery calculations.

If I have a queen and two rooks on a file (likely and desirable) or if I have a
bishop and two queens on a diagonal (rare in practice but desirable), then I
need to know about it.  I also want to know what is behind my ram for as deep as
I can batter.  E.g. does this queen sacrifice make sense?
[D]7k/3n2q1/5p2/8/6N1/2Q5/1B6/Q1K5 w - -




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.