I am reading the following paper 1998(H.Hudzik) P.574

It reads using L'Hopital rule$$\liminf_{u\to\infty} \frac{1/\varphi(1/u)}{\psi(u)}=\liminf_{u\to\infty}\frac{\varphi'(u)}{\psi'(u)u^2[\varphi(1/u)]^2}.$$ That means we can apply L'Hopital for lower limits i.e. $$\liminf_{u\to\infty} \frac{f(u)}{g(u)}=\liminf_{u\to\infty}\frac{f'(u)}{g'(u)}?$$ But I only know the classical one. Is there someone can give me some reference to check this formula? Or if possible someone can give a proof?


2 Answers 2


The full L'Hopital rule says that $$\liminf \frac{f'}{g'}\leq\liminf\frac{f}{g}\leq\limsup\frac{f}{g}\leq\limsup\frac{f'}{g'}.$$ So in the special case when the limit of $f'/g'$, exists, the limit of $f/g$ also exists and is equal to the limit of $f'/g'$.

This general rule is proved by integration.

  • 3
    $\begingroup$ For those interested, a little known paper of William Henry Young, On indeterminate forms [Proceedings of the London Mathematical Society (2) 8 (1910), pp. 40-76], extended this by proving (among other things) that every subsequential limit point of $f/g$ is a subsequential limit point of $f'/g'.$ In fact, I think Young's paper is the first paper in which even the weaker version involving $\liminf$ and $\limsup$ appears, the weaker version being (I think) rediscovered by Picone (1929) (continued) $\endgroup$ Jan 18 at 17:40
  • 2
    $\begingroup$ and Amante (1936) and Halvorsen (1963). A nice survey of many subtle versions of L'Hopital's rule is given in: Ioan Muntean, L'Hôpital's rules with extreme limits , pp. 11−28 in Seminar on Mathematical Analysis (Preprint 93-7), "Babeş-Bolyai" University, Cluj-Napoca, 1993 (MR1333230; Zbl 833.26007). Although I have a photocopy of this paper, I cannot find it on the internet. Muntean cites Picone's paper, but not the papers by Young or Amante or Halvorsen. $\endgroup$ Jan 18 at 17:57
  • $\begingroup$ Thank you very much for your nice reference. This is much helpful. $\endgroup$ Jan 19 at 8:05

Q: Is it true that $$\liminf_{u\to\infty} \frac{f(u)}{g(u)}\stackrel{?}{=}\liminf_{u\to\infty}\frac{f'(u)}{g'(u)}.$$ A: No, try $f(u)=u+\sin u$ and $g(u)=u$, then the left-hand-side of the equation equals 1, while the right-hand-side equals 0. The best you can do is replace = by $\geq$.

  • $\begingroup$ But my confused point why this paper use this rule, if we want this formula equal, which condition should we consider? $\endgroup$ Jan 18 at 10:55
  • $\begingroup$ the equality is correct if $\lim_{u\rightarrow\infty}f'(u)/g'(u)$ exists. $\endgroup$ Jan 18 at 11:29

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.