matlab - Match rows of 2 matrices for a minimum total sum of row-wise differences -

i have 2 matrices equal number of rows , columns. a(4x3) , b(4x3).

i match rows of rows of b when subtracting minimum sum. example of how inside of , b like:

a=  -0.3612   -0.1911   -15.3818      0.0376    0.0206    0.3498      0.0418    0.0229    0.3887      0.0188    0.0103    0.1749 

i'm doing this:

for j=1:size(b,1)     [vv, inds]=min(sum(abs(bsxfun(@minus,a,b(j,:))),2));     minjj(j,:)=vv;       indsmatch(j,:)=inds;     mm=min( minjj);      [a,ps] = removerows(a,'ind',inds);     [b,ps] = removerows(b,'ind',j);  end 

i error:

improper assignment rectangular empty matrix.

what i'm trying indsmatch. want 'order of rows indexes' of matrix paired rows of b (based on min difference). indexes of b kept 1 2 3 4.

here's vectorized approach. tries permutations of rows of b, , produces optimal row-permuted b in matrix bresult.

[m, n] = size(a); ind = perms(1:m);                                  % // permutations of row indices bb = permute(reshape(b(ind.',:),m,[],n),[1 3 2]);  %'// row-permuted b matrices  c = sum(sum(abs(bsxfun(@minus, a, bb)),2),1);      % // compute sum each [minsum, imin] = min(c);                           % // minimize bresult = b(ind(imin,:),:);                        % // output result