Mathematics Stack Exchange is a question and answer site for people studying math at any level and professionals in related fields. Join them; it only takes a minute:

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

Given $f:\mathbb{R}^{> 0} \rightarrow \mathbb{R}$ such that:

a) $f$ is $C^2$ in its domain.

b) $f$, $f'$, and $f''$ have finite limits as $x \rightarrow 0^+$

I want to extend $f$ to nonpositive values of $x$ (really just to a neighborhood of the origin) so that it remains $C^2$. Does the following construction work?

Let $\lim_{x\rightarrow 0^+} f = a$, $\lim_{x\rightarrow 0^+} f' = b$, $\lim_{x\rightarrow 0^+} f'' = c$, and define $g(x)$ to be equal to $f(x)$ if $x>0$ and to $a+bx+cx^2/2$ if $x \leq 0$.

In particular, how can I prove (if it is true) that $g$ as defined is continuously differentiable twice at $x=0$?

My motivation (you can ignore this part, but comments on it are welcome)

It seems to me that given a series $\sum_{n=1}^\infty a_n$ that happens to have the property that $a_n$ (currently a function of the positive integers) can be extended to a $C^2$ function $f(x)$ of $\mathbb{R}^{>0}$ at least for sufficiently high $x$, the convergence of the series should be able to be established by the behavior of $f$ and its first two derivatives "at infinity." More precisely, defining $f^*(x)=f(1/x)$, I think that if $f^*$ and its first derivative $\rightarrow 0$ as $x \rightarrow 0$, and its second derivative has a finite limit, then the series converges, and I am trying to prove this.

My idea for a proof is to show that $f^*$ is bounded by something of the form $kx^2$ for sufficiently small positive $x$. In this case, $\sum_{n=1}^\infty a_n = \sum_{n=1}^\infty f^*(1/n)$ would for sufficiently high $n$ be bounded term-by-term by $\sum k/n^2$ which converges. To prove this bound I would like access to Taylor's theorem at the origin, but I don't have this unless $f^*$ is actually twice-differentiable at the origin. This is why I want to create the extension I asked about above. If you have another or simpler route to my desired result (or a counterexample!) I'd be interested too.

share|cite|improve this question
up vote 2 down vote accepted

It may be worth considering the answer given here to a related question.

share|cite|improve this answer
It seems to me this fully answers the question. By your argument $g$ is differentiable with continuous derivative at the origin; then apply the result to $g'$ and we get what I want. Yes? – Ben Blum-Smith Aug 5 '11 at 13:01
Yes, that's the idea. – Shai Covo Aug 5 '11 at 13:06

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.