How to Move the Big O Notation from Denominator to Numerator📂Lemmas
How to Move the Big O Notation from Denominator to Numerator
Theorem
The following holds for a=0, p>0, and n∈N.
pa+O(hn)1=pa1+O(hn)
Explanation
It serves as a handy lemma for converting complex denominators into cleaner forms.
If there were no constant term a, it could neatly rise to pO(hn)1=O(h−pn) even without the lemma, but that is usually not very useful.
Proof
Let’s set q:=p1 for pa+O(hn)1=pa+bhn1=(a+bhn)−p1 and define the function f(h):=(a+bhn)−q.
When Maclaurin expanding for f,
f(h)=f(0)+hf′(0)+2h2f′′(0)+6h3f(3)(0)+⋯
Since f(h)=(a+bhn)−q,
f(0)=(a+b⋅0n)−q=pa1
Meanwhile,
f′(h)=−qbnhn−1(a+bhn)−q−1
Therefore, it’s f′(0)=0, and
f′′(h)=−qbn(n−1)hn−2(a+bhn)−q−1+bnhn−1(−q−1)bnhn−1(a+bhn)−q−2
Hence, it’s f′′(0)=0. Proceeding this way, when calculating the kth derivative f(k), we know that for k=2,3,⋯,(n−1) it’s f(k)(0)=0. For some g(h),
f(n)(h)=−qbn!hn−n(a+bhn)−q−1+hg(h)
Therefore,
f(n)(h)=−qbn!q+1a1
Thus,
f(h)=pa1+0+⋯+0−n!hnqbn!q+1a1+(n+1)!h(n+1)f(n+1)(0)+⋯
After 0, grouping all terms that appear later under hn, we obtain the following.
pa+O(hn)1=pa1+O(hn)