The idea is that we can always take a large enough multiple of \(b\) so that the remainder \(r\) is as small as possible. We do allow the possibility of \(r = 0\text{,}\) in which case we have \(b \mid a\text{.}\)