[Numpy-discussion] Distance Matrix speed

[Alex Cannon]

How about this?

def d5():
    return add.outer(sum(A*A, axis=1), sum(B*B, axis=1)) - \
                            2.*dot(A, transpose(B))