I'm given function $f:(a,\infty)\to\mathbb{R}$ which has a limit at infinity, i.e., $\lim_{x \to \infty}f(x)$ exists, call it $L$. And I want to show that given a function $g(x) := {f(1/x)},$ which is defined on $(0,1/a),$ that this function $g(x)$ has a limit at 0 if and only if the limit of $f$ as $x$ tends to infinity exists.

I know I have to use the $\epsilon - \delta$ defintion, but before that I think the following is an equivalent formulation: \begin{gather} \lim_{x \to \infty}f(x) = \lim_{x \to 0}f(1/x). \end{gather} I know this is just an exercise in chasing the $\epsilon - \delta$ notation, but I think the "trick" here is to use the fact that if $f$ has a limit at infinity, then for all $\epsilon > 0,$ there exists $M > a$ such that for all $x \geq M$ we have that $|f(x) - L| < \epsilon$. So I think the idea here is to pick my $\delta$ as $1/M$ since we have that \begin{gather} x \geq M \implies 1/x \leq 1/M \end{gather} and we know that if $x \geq M$ then $|f(x) - L| < \epsilon.$ So if we suppose $\epsilon_0 > 0$ and that $|f(1/x) - L| < \epsilon_0$ will $\delta_0 = 1/M$ suffice? My intuition says yes, but I am not sure how to formulate this rigorously.

  • 1
    $\begingroup$ Yes, this is quite a powerful trick that is often used to make certain limits more manageable to evaluate. Your reasoning for the validity of this technique is also pretty good! $\endgroup$ – Don Thousand Nov 29 '18 at 17:15
  • 1
    $\begingroup$ I should note, however, that the reverse is not the case. If $\lim_{x\to\infty}f(x)=a$, then $\lim_{x\to0}f(\frac1x)$ is not necessarily $a$. However, $\lim_{x\to0^+}f(\frac1x)=a$ $\endgroup$ – Don Thousand Nov 29 '18 at 17:24

Yes that's completely fine we have indeed

$$\lim_{y \to \infty}f(y) =L \iff \forall \epsilon>0 \quad \exists \bar y>0 \quad \forall y> \bar y \quad |f(y)-L|<\epsilon$$

and since for $g(x)= \frac1x$ we have

$$\lim_{x \to 0^+}g(x) =\infty \iff \forall M>0 \quad \exists \delta>0 \quad \forall x>0 \quad x<\delta \quad g(x)> M$$

then by $\bar y =M$ for $f(g(x))$ we have that

$$\forall \epsilon>0 \quad \exists \delta>0 \quad \forall x>0 \quad x<\delta \quad |f(g(x))-L|<\epsilon$$

that is

$$\lim_{x \to 0^+}f(g(x))=L$$


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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