[SOLVED] Infinite series, convergent or not?

**arbolis** · July 11th 2008, 11:50 AM

It has probably been answered before on MHF, but I didn't find it using the search button.
Tell whether the following infinite series is convergent or not :
$\sum_{n=2}^{+\infty} \frac{1}{n \ln (n)}$ . At first, I used the ration test, but I cannot conclude since it says "1". Then the integral test, which gave me divergent, but I'm not sure I did it well, because when I tried to get an approximation via Mathematica, I got an error like if it converges, it does it too slow or something like that. Nevertheless, it said an approximation is 13.4391. From it I doubted my integral test was well made. I tried the root test, but I'm not sure how to finish it.
Integral test : $\int_2^{+\infty} \frac{1}{x\ln (x)}$ . Let $u$ be $\ln (x)$ , then $du=\frac{1}{x}$ , and the integral becomes $\int_2^{+\infty} \frac{du}{u}=\ln u \big | ^{+\infty}_2$ . Replacing $u$ by $\ln (u)$ , it's easy to see the integral diverges.
So normally, so does the infinite series. Am I right?

**Jhevon** · July 11th 2008, 12:15 PM

Originally Posted by arbolis

It has probably been answered before on MHF, but I didn't find it using the search button.
Tell whether the following infinite series is convergent or not :
$\sum_{n=2}^{+\infty} \frac{1}{n \ln (n)}$ . At first, I used the ration test, but I cannot conclude since it says "1". Then the integral test, which gave me divergent, but I'm not sure I did it well, because when I tried to get an approximation via Mathematica, I got an error like if it converges, it does it too slow or something like that. Nevertheless, it said an approximation is 13.4391. From it I doubted my integral test was well made. I tried the root test, but I'm not sure how to finish it.
Integral test : $\int_2^{+\infty} \frac{1}{x\ln (x)}$ . Let $u$ be $\ln (x)$ , then $du=\frac{1}{x}$ , and the integral becomes $\int_2^{+\infty} \frac{du}{u}=\ln u \big | ^{+\infty}_2$ . Replacing $u$ by $\ln (u)$ , it's easy to see the integral diverges.
So normally, so does the infinite series. Am I right?

in terms of u, the limits are $\int_{\ln 2}^{\infty} \frac{du}{u}=\ln u \bigg | ^{\infty}_{\ln 2}$

yes, it is divergent

**Mathstud28** · July 11th 2008, 12:44 PM

Originally Posted by arbolis

It has probably been answered before on MHF, but I didn't find it using the search button.
Tell whether the following infinite series is convergent or not :
$\sum_{n=2}^{+\infty} \frac{1}{n \ln (n)}$ . At first, I used the ration test, but I cannot conclude since it says "1". Then the integral test, which gave me divergent, but I'm not sure I did it well, because when I tried to get an approximation via Mathematica, I got an error like if it converges, it does it too slow or something like that. Nevertheless, it said an approximation is 13.4391. From it I doubted my integral test was well made. I tried the root test, but I'm not sure how to finish it.
Integral test : $\int_2^{+\infty} \frac{1}{x\ln (x)}$ . Let $u$ be $\ln (x)$ , then $du=\frac{1}{x}$ , and the integral becomes $\int_2^{+\infty} \frac{du}{u}=\ln u \big | ^{+\infty}_2$ . Replacing $u$ by $\ln (u)$ , it's easy to see the integral diverges.
So normally, so does the infinite series. Am I right?

$\int_2^{\infty}\frac{dx}{x\ln^p(x)}$

Now lets look at the five cases

Case one

If $p<0$

$\int_2^{\infty}\frac{dx}{x\ln^p(x)}$ can be rewritten as $\int_2^{\infty}\frac{\ln^{-p}(x)}{x}dx$ where $-p>0\quad\text{because }p<0$

So then

$\int_2^{\infty}\frac{\ln^{-p}(x)}{x}dx>\int_2^{\infty}\frac{dx}{x}$

Therefore divergent.

Now what about when $p=0$

Obviously then we have

$\int\frac{dx}{x\ln^0(x)}=\int_1^{\infty}\frac{dx}{ x}=\infty$

So divergent.

How about when

$0<p<1$

We then have that

$\int\frac{dx}{x\ln^p(x)}dx=\frac{\ln^{-p+1}(x)}{-p+1}$

Now at $x=2\Rightarrow\frac{\ln^{-p+1}(2)}{-p+1}\text{ Which is a finite value}$

Now for $x\to\infty$ consider that since $0<p<1\Rightarrow{-p+1>0}$

And it can can be shown by direct substitution that $\lim_{x\to\infty}\ln^{n\in(0,1)}(x)=\infty$

Therefore divergent.

Now when $p=1$ we have that

$\int_2^{\infty}\frac{dx}{x\ln(x)}=\ln(\ln(x))\bigg |_2^{\infty}=\infty-\ln(\ln(2))$

So divergent.

Now how about when

$p>1$

we have that

$\int_2^{\infty}\frac{dx}{x\ln^p}=\frac{\ln^{-p+1}(x)}{-p+1}\bigg|_{2}^{\infty}$

Now consider that

$p>1\Rightarrow{-p+1<0}$

Now it can be shown very easily that $\lim_{x\to\infty}\ln^{n\in(-\infty,0)}(x)=0$

So then

$\int_2^{\infty}\frac{dx}{x\ln^p(x)}=0-\frac{\ln^{-p+1}(2)}{-p+1}$

Therefore convergent.

Putting this together we get

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

Converges only for $p>1$

**Moo** · July 16th 2008, 02:00 AM

Google for Bertrand series

Thread: [SOLVED] Infinite series, convergent or not?

LinkBack

Thread Tools

Display

[SOLVED] Infinite series, convergent or not?

Similar Math Help Forum Discussions

[SOLVED] For what value of p is this series convergent

Infinite Series, why divergent and not convergent?

Infinite Series: Divergent or Convergent?

Convergent infinite series

Infinite series - convergent vs divergent

Search Tags