MHB Finding the interval of convergence for a series with lnn

tmt1
Messages
230
Reaction score
0
So I have

$$\sum_{n = 2}^{\infty} \frac{1}{nln(n)}$$

I'm trying to apply the limit comparison test, so I can compare it to $b_n$ or $\frac{1}{n}$ and I can let $a_n = \frac{1}{nln(n)}$

Then I get $$\lim_{{n}\to{\infty}} \frac{n}{nln(n)}$$

Or $$\lim_{{n}\to{\infty}} \frac{1}{ln(n)}$$ Which is clearly 0.

Now, since the limit is 0, then the sums of series of $\sum_{ }^{}a_n$ would only converge if $\sum_{}^{}b_n$ converges. However $\sum_{}^{}b_n$ equals $\sum_{}^{}\frac{1}{n}$ which diverges.

The answer is that it is divergent, but I'm not sure how to prove it with my method. Is the method I am using not the right one to use, or am I misunderstanding the method?
 
Physics news on Phys.org
tmt said:
So I have

$$\sum_{n = 2}^{\infty} \frac{1}{nln(n)}$$

I'm trying to apply the limit comparison test, so I can compare it to $b_n$ or $\frac{1}{n}$ and I can let $a_n = \frac{1}{nln(n)}$

Then I get $$\lim_{{n}\to{\infty}} \frac{n}{nln(n)}$$

Or $$\lim_{{n}\to{\infty}} \frac{1}{ln(n)}$$ Which is clearly 0.

Now, since the limit is 0, then the sums of series of $\sum_{ }^{}a_n$ would only converge if $\sum_{}^{}b_n$ converges. However $\sum_{}^{}b_n$ equals $\sum_{}^{}\frac{1}{n}$ which diverges.

The answer is that it is divergent, but I'm not sure how to prove it with my method. Is the method I am using not the right one to use, or am I misunderstanding the method?

I would use the integral test.

First let $\displaystyle \begin{align*} f(x) = \frac{1}{x\ln{(x)}} \end{align*}$. We should note that $\displaystyle \begin{align*} f'(x) = -\frac{\left[ \ln{(x)} + 1 \right] }{\left[ x \ln{(x)} \right] ^2 } \end{align*}$ which is negative for all $\displaystyle \begin{align*} x \geq 2 \end{align*}$, and thus $\displaystyle \begin{align*} \frac{1}{x\ln{(x)}} \end{align*}$ is decreasing in this region.

As it's decreasing, that means $\displaystyle \begin{align*} \sum_{n = 2}^{\infty} \frac{1}{n\ln{(n)}} > \int_2^{\infty}{ \frac{1}{x\ln{(x)}}\,\mathrm{d}x } \end{align*}$, so evaluating the integral we have

$\displaystyle \begin{align*} \int_2^{\infty}{ \frac{1}{x\ln{(x)}}\,\mathrm{d}x } &= \int_{\ln{(2)}}^{\infty}{ \frac{1}{u}\,\mathrm{d}u } \textrm{ after substituting } u = \ln{(x)} \implies \mathrm{d}u = \frac{1}{x}\,\mathrm{d}x \\ &= \lim_{\epsilon \to \infty} \left[ \ln{|u|} \right] _{\ln{(2)}}^{\epsilon} \\ &= \lim_{\epsilon \to \infty} \ln{(\epsilon )} - \ln{\left[ \ln{(2)}\right] } \\ &\to \infty \end{align*}$

As the sum is greater than this infinite value, the sum also diverges.
 
For original Zeta function, ζ(s)=1+1/2^s+1/3^s+1/4^s+... =1+e^(-slog2)+e^(-slog3)+e^(-slog4)+... , Re(s)>1 Riemann extended the Zeta function to the region where s≠1 using analytical extension. New Zeta function is in the form of contour integration, which appears simple but is actually more inconvenient to analyze than the original Zeta function. The original Zeta function already contains all the information about the distribution of prime numbers. So we only handle with original Zeta...

Similar threads

Back
Top