Substract Matrix coloum-wise from Matrix without using a loop

Hello,

is there a way to substract all coloumns of one matrix one by one from another matrix without using a loop?

I am trying to calculate the minimum distance of all the 3D points in matrix B to all the points in matrix A, where A = [109x3]; B = [109x3].

WITH a loop I would do it that way:

for n=109:-1:1
    distance(n) = min(sqrt(sum((A - B(n,:)).^2,2)));
end

I was thinking of the repmat function but cant quite figure it out.

Thank your very much!

Answer 1

[Stephen23]

permute lets you do this on one line:

D = min(sqrt(sum(bsxfun(@minus,permute(A,[1,3,2]),permute(B,[3,1,2])).^2,3)),[],1);

And testing against your code (using two matrices of random values):

>> isequal(distance,D)
ans = 1

[kaju]

Thanks a lot!

Answer 2

[JESUS DAVID ARIZA ROYETH]

if you have Statistics and Machine Learning Toolbox :

distance=min(pdist2(A,B));

[kaju]

Sadly not, but thanks anyway!