span.m

Lattice coding and decoding

function vs = span(patt,dim)

%SPAN Linear combinations spanned over N-dimensional vector space
%   VS = SPAN(PATT,DIM) expands the pattern patt and creates all the
%   linear combinations over the $n$-dimensional vector space. 
%
%   Note: Such combinations could be also obtained via standard MATLAB
%         functions (see the code below). But there are certain
%         restrictions you should know about.  PERMS function is
%         practically usable only when its argument is less that about 10.
%         Furthermore for higher dimensions the computation via PERMS
%         function is too slow and moreover it doesn't produce the same
%         result as the SPAN function.
%
%         vs = span([0:2],3)
%         x = perms(0:2);
%         [m,n] = size(x);
%         k = m * n / 3; 
%         x = reshape(x,[3 k]); 
%         x = sortrows(x');
%
%   See also BLUR, CLUSTER.

%   Copyright 2001-2003 Kamil Anis, [e-mail]
%   Dept. of Radioelectronics, 
%   Faculty of Electrical Engineering
%   Czech Technical University in Prague
%   $Revision: 0.1 $  $Date: 2003/1/16 17:33:28 $
%   --
%   <additional stuff goes here>

m = length(patt);

for i = 1:dim
    k = abs(i - dim);
    l = i - 1;
    z = cluster(patt',m ^ k,m ^ l);
    vs(:,i) = z;
end