Zeta pushed to the limit 3

Calculus Level 5

lim n ζ ( H n ) ζ ( ln n ) \lim_{n \to \infty} \frac{\zeta'(H_n)}{\zeta'(\ln n)}

Evaluate the limit above, where ζ ( s ) = d d s ζ ( s ) \zeta'(s) = \dfrac{d}{ds} \zeta(s) .


You might also enjoy part 1 and/or 2 .


The answer is 0.67025609338.

This section requires Javascript.
You are seeing this because something didn't load right. We suggest you, (a) try refreshing the page, (b) enabling javascript if it is disabled on your browser and, finally, (c) loading the non-javascript version of this page . We're sorry about the hassle.

1 solution

Julian Poon
Nov 5, 2015

ζ ( s ) = k = 1 ln ( k ) k s \zeta '(s)=-\sum _{ k=1 }^{ \infty }{ \frac { \ln { (k) } }{ { k }^{ s } } }

Since as n n approaches infinity, H n = k + ln n H_{n}=k+\ln{n} , where k k is the Euler–Mascheroni constant, lim n ζ ( H n ) ζ ( ln n ) = lim n ζ ( ln n + k ) ζ ( ln n ) = lim n ζ ( n + k ) ζ ( n ) \lim _{ n\to \infty } \frac { \zeta '(H_{ n }) }{ \zeta '(\ln { n } ) } =\lim _{ n\to \infty } \frac { \zeta '(\ln { n } +k) }{ \zeta '(\ln { n } ) } =\lim _{ n\to \infty } \frac { \zeta '(n+k) }{ \zeta '(n) }

= ln ( 2 ) 2 n + k + ln ( 3 ) 3 n + k . . . ln ( 2 ) 2 n + ln ( 3 ) 3 n . . . = ln ( 2 ) + ln ( 3 ) 2 n + k 3 n + k . . . ln ( 2 ) 2 k + ln ( 3 ) 2 n + k 3 n . . . = 2 k =\frac { \frac { \ln { \left( 2 \right) } }{ 2^{ n+k } } +\frac { \ln { \left( 3 \right) } }{ 3^{ n+k } } ... }{ \frac { \ln { \left( 2 \right) } }{ 2^{ n } } +\frac { \ln { \left( 3 \right) } }{ 3^{ n } } ... } =\frac { \ln { \left( 2 \right) } +\frac { \ln { \left( 3 \right) } 2^{ n+k } }{ 3^{ n+k } } ... }{ \ln { \left( 2 \right) } 2^{ k }+\frac { \ln { \left( 3 \right) } 2^{ n+k } }{ 3^{ n } } ... } =2^{ -k }

Hence lim n ζ ( H n ) ζ ( ln n ) = 2 k = 0.6702560933... \lim _{ n\to \infty } \frac { \zeta '(H_{ n }) }{ \zeta '(\ln { n } ) } =2^{ -k }=\boxed{0.6702560933...}

Moderator note:

There are several issues with this solution:

  1. It is not true that H n = k + ln n H_n = k + \ln n . What we have is that \lim_ H_n - \ln n = k .
  2. For a continuous function f f , and two sequences a n , b n a_n, b_n with lim a n b n = 0 \lim a_n - b_n = 0 , it is not true that lim f ( a n ) f ( b n ) = 0 \lim f(a_n) - f(b_n) = 0 . It is true in the case where lim a n \lim a_n is finite, but with a n = H n a_n = H_n , the limit is infinity. As a simple counter example, consider a n = n , b n = n + 1 n a_n = n, b_n = n + \frac{1}{n} and f ( x ) = x 2 f(x) = x^2 . We have lim f ( a n ) f ( b n ) = 2 \lim f(a_n) - f(b_n) = 2 .

As such, the second line hasn't been justified.

Actually, ζ ( s ) = n = 1 ln n n s \displaystyle \zeta'(s) = -\sum_{n=1}^{\infty} \frac{\ln n}{n^s} , so you're missing a negative sign.

Alternatively, one can show that lim ζ ( a ) ζ ( b ) = lim ζ ( a ) 1 ζ ( b ) 1 \lim \frac{\zeta'(a)}{\zeta'(b)} = \lim \frac{\zeta(a)-1}{\zeta(b)-1} by l'Hopital's rule.

Jake Lai - 5 years, 7 months ago

Log in to reply

I too realised that and made a problem on it.

Julian Poon - 5 years, 7 months ago

Issue 1: Since in the question, we are considering lim n H n \lim_{n\to \infty} H_{n} , it should be okay to use lim n H n = ln ( n ) + k \lim_{n\to \infty} H_{n}=\ln(n)+k .

Issue 2: Yeah, that part is wrong. Opps. The answer is still 2 k 2^{-k} though since in the later steps, sticking to ln ( n ) \ln(n) still makes the limit converge to 2 k 2^{-k} .

Julian Poon - 5 years, 6 months ago

0 pending reports

×

Problem Loading...

Note Loading...

Set Loading...