wolffd@0: function sub = ind2subvKPM(siz, ndx)
wolffd@0: % IND2SUBV Like the built-in ind2sub, but returns the answer as a row vector.
wolffd@0: % sub = ind2subv(siz, ndx)
wolffd@0: %
wolffd@0: % siz and ndx can be row or column vectors.
wolffd@0: % sub will be of size length(ndx) * length(siz).
wolffd@0: %
wolffd@0: % Example
wolffd@0: % ind2subv([2 2 2], 1:8) returns
wolffd@0: %  [1 1 1
wolffd@0: %   2 1 1
wolffd@0: %   ...
wolffd@0: %   2 2 2]
wolffd@0: % That is, the leftmost digit toggle fastest.
wolffd@0: %
wolffd@0: % See also SUBV2IND
wolffd@0: 
wolffd@0: n = length(siz);
wolffd@0: 
wolffd@0: 
wolffd@0: if  n==0
wolffd@0:   sub = ndx;
wolffd@0:   return;
wolffd@0: end  
wolffd@0: 
wolffd@0: if all(siz==2)
wolffd@0:   sub = dec2bitv(ndx-1, n);
wolffd@0:   sub = sub(:,n:-1:1)+1;
wolffd@0:   return;
wolffd@0: end
wolffd@0: 
wolffd@0: cp = [1 cumprod(siz(:)')];
wolffd@0: ndx = ndx(:) - 1;
wolffd@0: sub = zeros(length(ndx), n);
wolffd@0: for i = n:-1:1 % i'th digit
wolffd@0:   sub(:,i) = floor(ndx/cp(i))+1;
wolffd@0:   ndx = rem(ndx,cp(i));
wolffd@0: end
wolffd@0: 
wolffd@0: 
wolffd@0: 
wolffd@0: %%%%%%%%%%
wolffd@0: 
wolffd@0: function bits = dec2bitv(d,n)
wolffd@0: % DEC2BITV Convert a decimal integer to a bit vector.
wolffd@0: % bits = dec2bitv(d,n) is just like the built-in dec2bin, except the answer is a vector, not a string.
wolffd@0: % n is an optional minimum length on the bit vector.
wolffd@0: % If d is a vector,  each row of the output array will be a bit vector.
wolffd@0: 
wolffd@0: 
wolffd@0: if (nargin<2)
wolffd@0:   n=1; % Need at least one digit even for 0.
wolffd@0: end
wolffd@0: d = d(:);
wolffd@0: 
wolffd@0: [f,e]=log2(max(d)); % How many digits do we need to represent the numbers?
wolffd@0: bits=rem(floor(d*pow2(1-max(n,e):0)),2);