[jianpan]

On book page 44, the growth function of positive intervals is defined as 2C(N+1). This is the fomular of selection without replacement, which implies we can't pick the same interval twice. But the book says " if the both end values fall in the same region, ....", which means we do allow to pick the same interval twice. If that's case, we should use selection with replacement N^2 in this case. This doesn't change the order of polynomial, but seems more rigorous.