Find minimum difference between matrices.

I have two matrices. For each element in the first matrix, I want to find the closest value in the second matrix, and then take the difference, which will be recorded in a third matrix. Any thoughts on the best way to do this? Thank you.

Answer 1

[Azzi Abdelmalek]

A=randi(4,4)
B=randi(4,4)% Example
b=B(:);
for k=1:numel(A)
  [val,idx]=min(abs(b-A(k)));
  C(k)=val;
end
C=reshape(C,size(A))

%If you do not allow repetition

A=randi(4,4)
B=randi(4,4)
b=B(:);
for k=1:numel(A)
  [val,idx]=min(abs(b-A(k)));
  C(k)=val;
  b(idx)=[];
end
C=reshape(C,size(A))

Answer 2

[Matt Kindig]

Another way using histc() function (may be faster for large matrices).

A = rand(5,5);  %first matrix
B = rand(2,3);   %second matrix
B = sort(B(:));     %make B vector and sort for binning convenience
edges = B(1:(end-1)) + 0.5*diff(B);   %"edges" are halfway between each B span
edges= [-inf; edges; inf];   %add +/- inf at ends to include all values
[~,whichBin] = histc(A, edges);   %get which bin contains each A
% for each value in A, the value in B which is closest to A 
%will fall between bin edges k and k+1 (i.e., k-th element of B)
%third matrix is difference between A and the B for that bin
DiffAB = A - B(whichBin);

EDIT: changed to allow A and B to be matrices as well (consistent with Azzi's solution).