1
$\begingroup$

As the title says the task is to find the limit:$\\$

$$\lim_{n\to\infty}\frac{\ln2^{1/n}-\ln n^2}{1+1/2+1/3+...+1/n})$$

$\\$ I assume I should attempt to somehow reach the form of $\lim_{n\to0}\frac{\ln(1+n)}{n}$ or use the Cesaro-Stolz theorem,but both of those attempts of mine have failed to get me the correct answer. I would appreciate any kind of hints. And also, I'm not allowed to use L'hospital's rule, or any kind of derivation at that.

Thank you in advance!

EDIT: I am not to use integrals either.

2 Answers 2

2

It is easy to see that the first fraction limits to zero, it is just $$\frac{\ln 2}{nH_n}.$$ Thus the problem boils down to showing that $$\frac{\ln n}{H_n}\to 1$$ This is either well known or follows from Stolz theorem since $$(n+1)\ln (1+\frac{1}{n})\to 1$$

  • 0
    What "first fraction"? As far as I can see, the limit is $\;-2\;$ ...2017-01-20
  • 0
    @DonAntonio Yes thats what my answer shows, -2 is the limit.2017-01-20
  • 0
    @ReneSchipperus Ok, thank you. I'll take it from you...and, anyway, the OP seems to have understood (because I didn't). I still can't see what "the first fraction" is...2017-01-20
  • 0
    @DonAntonio I'm just splitting the numerator to make it a sum of two fractions.2017-01-20
2

Use some quick log rules to see that

$$\ln(2^{1/n})-\ln(n^2)=\frac1n\ln(2)-2\ln(n)$$

Also note that

$$\ln(n)=\int_1^n\frac1x\ dx\le\sum_{k=1}^n\frac1k\le1+\int_1^n\frac1x\ dx=1+\ln(n)$$

Thus,

$$\frac{\frac1n\ln(2)-2\ln(n)}{\ln(n)}\le\frac{\ln(2^{1/n})-\ln(n^2)}{\sum_{k=1}^n\frac1k}\le\frac{\frac1n\ln(2)-2\ln(n)}{1+\ln(n)}$$

And apply the squeeze theorem (divide all fractions by $\ln(n)$ if you still need help)

  • 0
    I'm sorry I haven't mentioned that I don't know how to use integrals yet honestly, and even if I did I am not allowed to, I will edit my question accordingly!2017-01-20
  • 0
    @DomoB Ah darn, this was by far the simplest route in my eyes, but oh well :D2017-01-20
  • 0
    Thanks a lot for your answer! And sorry.. x)2017-01-20
  • 0
    @DomoB :P No problem. I suppose you could say domo arigato :32017-01-20
  • 0
    @SimpleArt Nice idea, yet the denominators in both extremes in your last line are interchanged as everything positive here: on the left it must be $\;1+\log n\;$ , and on the right one $\;\log n\;$2017-01-20
  • 0
    @DonAntonio I was pretty sure the numerator is negative for large $n$.2017-01-20
  • 0
    @SimpleArt Rats, I forgot about the numerators! You're right...:) +12017-01-20
  • 0
    @DonAntonio Same. I actually had it flipped the way you mentioned, and then I saw it the same, since I realized the limit came out to be negative two...2017-01-20
  • 0
    @ParamanandSingh thanks :-)2017-01-21