Suppose you have two (nice enough) functions $f$ and $g$ and a constant $\lambda$ such that $$\lim_{x\to\infty}\frac{f(x)}{g(x)}=\lambda$$ Is it true that $$\lim_{x\to\infty}\frac{f^{-1}(x)}{g^{-1}(x/\lambda)}=1$$

The "reasoning" goes like this: $$\frac{f(x)}{g(x)}\approx\lambda$$ $$f(x)\approx\lambda g(x)$$ $$x\approx f^{-1}(\lambda g(x))$$ $$g^{-1}(x/\lambda)\approx f^{-1}(x)$$ $$\frac{f^{-1}(x)}{g^{-1}(x/\lambda)}\approx 1$$

all of this supposing there is no problem in $x\mapsto f^{-1}(x)$ and $x\mapsto g^{-1}(x/\lambda)$

I think assuming "nice enough" (continuity, inverse, ...) and being a bit more precise like $$f(x)=\lambda g(x)+o(g(x))$$ will prove the statement.

What I'm more interested in is under what conditions does it remain true in discrete variable (and non-existing inverse function for $f$ or $g$ or both)


  • $\begingroup$ Why do you have $n$ in the limit? $\endgroup$ Aug 18 '18 at 23:28
  • 6
    $\begingroup$ Where does $g^{-1}(x/\lambda)\approx f^{-1}(x)$ come from? $\endgroup$
    – Jack M
    Aug 19 '18 at 0:36
  • $\begingroup$ @ThomasAndrews thanks! i wanted to write $x$ $\endgroup$ Aug 19 '18 at 1:05
  • $\begingroup$ @JackM from setting $x\mapsto g^{-1}(x/\lambda )$ $\endgroup$ Aug 19 '18 at 1:06
  • $\begingroup$ @Pedro I don't understand what that means. We have $g(x)\approx \frac{f(x)}{\lambda}$, but not $g(x)\approx\frac x\lambda$ $\endgroup$
    – Jack M
    Aug 19 '18 at 10:35

No. If you take $f(x) = \log(x)$ and $g(x) = \log(x)-1$ and $\lambda = 1$ then :

$$ \underset{x\rightarrow\infty}{\lim}\frac{f(x)}{g(x)} = 1 $$

But :

$$ \frac{f^{-1}(x)}{g^{-1}(x)} = \frac{e^x}{e^{x+1}} = e^{-1} \neq 1$$

  • $\begingroup$ Nice counterexample! Thanks $\endgroup$ Aug 19 '18 at 1:10

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.