> otherwise, there is an 
> obvious "mathematically guaranteed way of solving the puzzles" 
> in principle.  Is there not?

Yes, there is.  

Let P be a Sudoku puzzle specified as a 9 9 matrix 
with the blank squares being 0.  Then
   u=: 1+9 9$"1 (#:i.@(*/)) 9^81x 
generates all possible 9 9 matrices of the integers from 1 to 9;
   u=: (#~ ((0=P)+.P&=)"2) u
selects matrices that have the specified digits in P in
the correct places;
   u=: (#~ f ) u 
selects the matrices with all 9 digits in each row, where 
   f=: (9$1) -:"1 */@((1+i.9)&e.)"1
and 
   u=: (#~ f@:(|:"2)) u
selects the matrices with all 9 digits in each column, and
   u=: (#~ f@:(,/@((;~9$1 0 0)&(,;.1))"2)) u
are all the solutions for P.



----- Original Message -----
From: Jose Mario Quintana <[email protected]>
Date: Tuesday, March 17, 2009 11:52
Subject: [Jgeneral] [JGeneral] The first mathematically guaranteed way of 
solving Sudoku puzzles?
To: [email protected]

> Yesterday I read an article in USA TODAY about "the first 
> mathematically guaranteed way of solving the puzzles;" see 
> http://www.usatoday.com/tech/science/mathscience/2009-03-15-
> sudoku-secret_N.htm.  It reminded me of an interesting old forum 
> thread: http://www.jsoftware.com/pipermail/programming/2005-December/thread.html#298 (see
>  also http://www.vector.org.uk/archive/v214/sudoku.htm) and a hint from 
> an academic female adviser "It is not just about laying the eggs; you must 
> know how to cluck them."
> 
> In any case, it seems to me that the 'way' should be qualified, 
> in some sense, as 'implementable;' otherwise, there is an 
> obvious "mathematically guaranteed way of solving the puzzles" 
> in principle.  Is there not?
----------------------------------------------------------------------
For information about J forums see http://www.jsoftware.com/forums.htm

Reply via email to