16

Let $f:(0,\infty) \to R$ be differentiable. Suppose that $\lim_{x\to\infty}(f(x)+f'(x))=L$. Show that $\lim_{x\to\infty} f(x) = L$ and $\lim_{x\to\infty} f'(x) = 0$. (Hint: Write $f(x) = e^xf(x)/e^x$ and use l’Hopital’s Rule.)

My working for $\lim_{x\to\infty}f'(x)=0$:

For $\lim_{x\to\infty}f'(x) = 0$, I let $f(x) = e^xf(x)/e^x$ and applied quotient rule which then cancels off $e^{2x}$ and I'm left with $\lim_{x\to\infty}(f(x)+f'(x)-f(x))$. Can I then equate this with $\lim_{x\to\infty} \left(f(x)+f'(x)\right) - \lim_{x\to\infty}f(x)$ which then gives $L-L=0$? Is this step correct?

user26857
  • 52,094
eee
  • 229
  • 5
    Hi @eee! $$\color{red}{\Large\text{Welcome to Math.SE!}}$$ Don't worry about it now (since you're new) but you might like to know that we use MathJax here (e.g. $\theta$ for $\theta$). Also, please show your working so far :) – Shaun Jan 23 '14 at 11:50
  • 1
    Well, did you try computing $\lim_{x\rightarrow\infty} f(x)$ using the Hint (Note, L'Hôpital's rule can be used since the denominator of $e^xf(x)/e^x$ tends to $\infty$ as $x\rightarrow\infty$)? – David Mitra Jan 23 '14 at 11:51
  • i am stuck at getting limx→∞f(x).. – eee Jan 23 '14 at 12:07
  • 4
    $$\lim_{x\rightarrow\infty}f(x)=\lim_{x\rightarrow\infty}{e^xf(x)\over e^x}\ \buildrel{\text LH}\over= \ \lim_{x\rightarrow\infty}{\bigl(e^xf(x)\bigr)'\over (e^x)' }=\lim_{x\rightarrow\infty}{e^xf(x)+e^x f'(x)\over e^x}=\lim_{x\rightarrow\infty}\bigl(f(x)+f'(x)\bigr).$$ – David Mitra Jan 23 '14 at 12:12
  • OH I made a careless mistake product rule involves + and not - how careless can i be! thanks a lot david :) – eee Jan 23 '14 at 12:14
  • @David Mitra: For your argument you have to assume that $L\ne 0$, take for example $f(x)=e^{-x}$ – gammatester Jan 23 '14 at 12:17
  • @gammatester I don't see your point... – David Mitra Jan 23 '14 at 12:21
  • i dont think there's a need to assume L≠ 0.. – eee Jan 23 '14 at 12:22
  • @David Mitra: You must have $\lim_{x\rightarrow \infty}e^x f(x) = \infty,$ to apply L'Hôpital in the given form, which is not the case for $f(x)=e^{-x},,$ i.e. $L=0$. – gammatester Jan 23 '14 at 12:29
  • @DavidMitra for lim f′(x) = 0, i used quotient rule which then cancels off e^2x and i'm left with lim(f(x)+f'(x)-f(x)). can i then equate this with lim(f(x)+f'(x))-lim(f(x)) which then gives L-L=0? is this step correct? – eee Jan 23 '14 at 12:33
  • 2
    @gammatester No. The rule applies whenever the denominator has infinite limit. It does not matter what the numerator is doing (the limit of the numerator need not even exist). This is mentioned on the Wiki page as a note in the "General proof" section, as well as several Analysis texts. E.g., Bartle and Sherberts'. (For some unfathomable reason, this form is not mentioned in most calculus texts.) – David Mitra Jan 23 '14 at 12:42
  • @David Mitra: I do not doubt your result, but the Wiki pages and e.g. http://planetmath.org/lhopitalsrule or http://www.encyclopediaofmath.org/index.php?title=L%27Hospital_rule&oldid=14236 all assume the form $\frac{0}{0}$ or $\frac{\infty}{\infty}$. I will look into some off-line sources. There is no need to continue this discussion. – gammatester Jan 23 '14 at 12:45
  • I have difficulty proving the second part of the question where lim f'(x)=0, and I think my working above is wrong. kindly enlighten me thank you. :) – eee Jan 23 '14 at 12:46
  • 2
    @gammatester See Theorem 7.1, page 127, from these notes of Pete L. Clark. The proof there is taken from baby Rudin. – David Mitra Jan 23 '14 at 12:49
  • @David Mitra: Thank you. Meanwhile I found the mention note on the Wiki page too. – gammatester Jan 23 '14 at 12:56
  • Hmm,I'm still stuck at proving the part where limf'(x)=0, my working seemed 'illegal', any headstart? – eee Jan 23 '14 at 14:07

2 Answers2

6

You have function $f$, and it's given that $\lim_{x \to \infty} (f(x) + f'(x)) = L.$

We'll assume that $\lim_{x \to \infty} f'(x) = 0.$

If $\lim_{x \to \infty} f'(x) = 0.$, then we can rewrite:

$$\lim_{x \to \infty} (f(x) + f'(x)) = L$$ as $$\lim_{x \to \infty} (f(x) + 0) = \lim_{x \to \infty} f(x) = L.$$

If we'll prove that $\lim_{x \to \infty} f(x) = \lim_{x \to \infty} (f(x) + f'(x))$, then we've proved that $f'(x) = 0$.

Using the hint you gave, we'll turn $f(x)$ into $\frac{e^x f(x)}{e^x}$ and using LHopital's rule that says:

$$\lim_{x \to \infty} \frac{g(x)}{h(x)} = \lim_{x \to \infty} \frac{g'(x)}{h'(x)}$$.

In our case, $$g(x) = {e^xf(x)} \text{ and } h(x) = e^x. $$

Let's start:

$$1. \text{Turn to suggested form:} \lim_{x \to \infty} f(x) = \lim_{x \to \infty} \frac{e^x f(x)}{e^x}.$$

$$2. \text{Apply LHopital's rule: differentiate both numerator and denumerator}:$$

$$ \lim \frac{(e^xf(x))'}{(e^x)'} = \lim = \frac{e^x f(x) + e^x f'x}{e^x} = \lim (f(x) + f'(x)).$$

We've got that $f(x) = f(x) + f'(x)$ that meets out assumation.

Therefore, $f'(x) = 0$.

Proven.

Billie
  • 3,449
  • Well you have to prove that $f'(x)=0$... – Umberto Jan 23 '14 at 14:12
  • @user1798362 sorry I don't really get your point.. – eee Jan 23 '14 at 14:16
  • @Umberto: But in this case i cannot use definition of derivative correct? because x tends to infinity.. – eee Jan 23 '14 at 14:18
  • 1
    Well you are actually (altough not explicitly) working with two limits. One in your derivative definition and one for $x$ going to infinity. The hint given is very useful, but mathematically speaking, an ugly way of solving the problem (estetically speaking...). – Umberto Jan 23 '14 at 14:23
  • @eee I have edited my answer. – Billie Jan 23 '14 at 14:28
  • @user1798362: did you include the bracket? the question is lim (f(x)+f'(x)) = L. did you mean the same thing in your first line? – eee Jan 23 '14 at 14:31
  • @eee It doesn't matter when there's only + and no $\cdot$ or these things. the answer is the same for $\lim_{x \to \infty} (f(x) + f'(x)$ – Billie Jan 23 '14 at 14:33
  • oh ok so are you using lim(f(x)+g(x))=lim f(x) + lim g(x) ? – eee Jan 23 '14 at 14:34
  • @eee this is my write style. I edited the post using the brakets. I meant exactly to the expression you asked. – Billie Jan 23 '14 at 14:35
  • ok i used the same working to prove the first part of the question limf(x)=L so am i right to say if i have derived the first part, i can easily obtain the second part? – eee Jan 23 '14 at 14:38
  • @eee You are right – Billie Jan 23 '14 at 14:40
  • L = lim (f(x)+f'(x)) = lim(f(x)) + lim(f'(x)) = L + 0, so lim(f'(x))=0? – eee Jan 23 '14 at 14:42
  • @user1798362 it really is that easy to prove the second part? thank you! – eee Jan 23 '14 at 14:50
  • @eee Yes, this is a basic question .. If I helped you, please accept my answer – Billie Jan 23 '14 at 15:04
  • is it just press tick will do? sorry i'm new. :) – eee Jan 23 '14 at 15:04
  • @eee Yes :) I glad to be able to help you, can't forget the days I needed this help.. – Billie Jan 23 '14 at 15:05
  • oh haha thanks a lot! anw actually i have one more question halfway done, do you mind? it's here:http://math.stackexchange.com/questions/648730/fx-cos-x-prove-that-f-is-differentiable-at-these-points-and-not-differe/648744?noredirect=1#648744 – eee Jan 23 '14 at 15:06
  • @user1798362: um does this stackexchange allow us to add friends? how do i approach you next time if i have a question? :) – eee Jan 23 '14 at 15:14
  • @eee I'm not as good in Calculus as you think .. it's just a very basic question .. – Billie Jan 25 '14 at 11:06
0

In case some doubt the condition of application of L'Hopital rule, we can use the discreted substitution version - i.e. Stolz theorem,which states that for two sequences $\{a_n\}$ and $\{b_n\}$, if $b_n\rightarrow\infty$ monotonously,then as long as $\lim_{n\rightarrow\infty} \frac{a_n-a_{n-1}}{b_n-b_{n-1}}$ exists (could be $\pm \infty$),we have \begin{align*} \lim_{n\rightarrow\infty} \frac{a_n-a_{n-1}}{b_n-b_{n-1}}=\lim_{n\rightarrow\infty} \frac{a_n}{b_n}. \end{align*} Now suppose $\lim_{x\rightarrow \infty} f(x)$ is not $L$,then there exists $M>0$, such that for each $X$, there exists $x>X$ such that $|f(x)-L|>M$.Choose monotonous sequence $\{x_n\}\rightarrow \infty$ such that $|f(x_n)-L|>M$.Consider $a_n=e^{x_n}(f(x_n)-L)$ and $b_n=e^{x_n}$. Then \begin{align*} \lim_{n\rightarrow\infty} \frac{a_n-a_{n-1}}{b_n-b_{n-1}}&=\lim_{n\rightarrow\infty} \frac{e^{x_n}(f(x_n)-L)-e^{x_{n-1}}(f(x_{n-1})-L)}{e^{x_n}-e^{x_{n-1}}}\\ &=\lim_{n\rightarrow\infty}\frac{e^{\xi_n}(f(\xi_n)+f'(\xi_n)-L)}{e^{\xi_n}}\\ &=\lim_{n\rightarrow\infty} f(\xi_n)+f'(\xi_n)-L\\ &=0. \end{align*} So by Stolz theorem we see \begin{align*} \lim_{n\rightarrow\infty} \frac{a_n}{b_n}&=\lim_{n\rightarrow\infty} \frac{e^{x_n}(f(x_n)-L)}{e^{x_n}}\\ &=\lim_{n\rightarrow\infty}f(x_n)-L\\ &=0, \end{align*} which contradicts that $|f(x_n)-L|$ has a non-zero lower bound. This may be equivalent to the proof of L'Hopital as the Stolz does, but I think it may resolve the doubt about where L'Hopital rule can apply.

Egyptian
  • 163
  • 7