1
$\begingroup$

Does the series $\frac1{n^n}$ converges or diverges .

By comparison test I can claim that $\frac1{n^n} < \frac1n$ , and since series $\frac1n$ appears divergent , so the series $\frac1{n^n}$ also diverges. However by Cauchy root test the limit of $\frac1{n^n}$ appears to be $0<1$ which suggests the convergence of the series

$\endgroup$
  • 2
    $\begingroup$ A series of nonnegative terms with all terms greater than or equal to a divergent series diverges. To dramatize your error, consider the series with all terms equal to zero. By your logic, since $0 < 1/n$ for all $n$, it would diverge. $\endgroup$ – quasi Oct 2 '17 at 9:46
  • 1
    $\begingroup$ Your argument would be valid if $1/n^n \color{red}{>} 1/n$. But it is not. $\endgroup$ – M. Winter Oct 2 '17 at 9:50
  • $\begingroup$ Series $\dfrac{1}{n^n}$ converges so quickly that just adding $5$ terms gives the result with $4$ exact decimals. Adding $10$ terms gives $1.29128\,599706$ which has eleven exact decimals $\endgroup$ – Raffaele Oct 2 '17 at 10:06
  • $\begingroup$ @Raffaele: quite right, anyway shouldn't leave the OP the impression that this is a proof. What about $1/n^{n(20-n)}$ ? $\endgroup$ – Yves Daoust Oct 2 '17 at 10:14
2
$\begingroup$

We cannot conclude that since $\frac{1}{n^n} < \frac{1}{n}$ then $\sum_{n=1}^\infty \frac{1}{n^n}$ diverges.

If $\frac{1}{n^n} > \frac{1}{n}$ (which is not true), then we can conclude that $\sum_{n=1}^\infty \frac{1}{n^n}$ diverges.

You have used Cauchy root test to conclude that it converges.

If we want to use comparison test, notice that $$\frac{1}{n^n} \leq\frac{1}{n^2}$$

and since $\sum_{n=1}^\infty \frac1{n^2}$ conveges, hence $\sum_{n=1}^\infty \frac1{n^n}$ converges.

$\endgroup$
  • $\begingroup$ Then is it that this given series converges by Cauchy root test ? $\endgroup$ – Nabendra Oct 2 '17 at 9:50
  • $\begingroup$ yes, you have proven it using the root test. $\endgroup$ – Siong Thye Goh Oct 2 '17 at 9:51
  • $\begingroup$ Now suppose if I use Ratio Test then, let u(n)=1/n^n and so u(n+1)=1/(n+1)^(n+1). Hence u(n)/u(n+1)=(1+1/n)^n * (n+1) and thus taking limit n tends to infinity on the above I get infinity(not sure whether calculations are correct) , which stands as greater than 1. So can I claim , the series in convergent by Ratio Test. $\endgroup$ – Nabendra Oct 2 '17 at 9:56
  • $\begingroup$ Usually for ratio test, we compute $\left| \frac{u(n+1)}{u(n)}\right|$ and show that it is less than $1$. $\endgroup$ – Siong Thye Goh Oct 2 '17 at 10:28
3
$\begingroup$

The comparison test tells you that if $$a_n> b_n$$ and $$\sum_{n}a_n$$ converges, then $$\sum_n b_n$$ also converges.

It does not say that if $\sum_{n} a_n$ diverges then $\sum_{n} b_n$ diverges. If it did, then because $\frac{1}{2^n} < 1$, you could conclude that $\sum \frac1{2^n}$ also diverges.

In fact, you would need the inequality reversed, so if $a_n<b_n$ and $\sum a_n$ diverges, then $\sum b_n$ also diverges.

So, your logic is incorrect.


In fact, the series $\sum\frac{1}{n^n}$ converges by the comparison test with $\frac{1}{2^n}$ or with $\frac{1}{n^2}$.

$\endgroup$
  • $\begingroup$ Does it appear convergent or divergent ? How gonna I check it ? $\endgroup$ – Nabendra Oct 2 '17 at 9:57
  • $\begingroup$ @Nabendra I really really dislike users that don't read my answers before they ask questions about them. $\endgroup$ – 5xum Oct 2 '17 at 9:57
  • $\begingroup$ I didn't understood your answer , as if I use 1/n the it diverges by comparison test instead of 1/n^2 or 1/2^n $\endgroup$ – Nabendra Oct 2 '17 at 10:04
  • $\begingroup$ @Nabendra You didn't understand my answer because you didn't bother reading it to the end. So I don't see why I should bother helping you any further. $\endgroup$ – 5xum Oct 2 '17 at 10:04
  • $\begingroup$ I do apologise, for my deed $\endgroup$ – Nabendra Oct 2 '17 at 10:07
0
$\begingroup$

A series converges if it has a converging upper bound and it diverges if it has a diverging lower bound. In other cases, you cannot conclude.

In this particular case, you can for instance exploit

$$\frac1{n^n}\le \frac1{2^n}$$ for $n\ge2$. Actually, the sequence converges extremely quickly, super-exponentially.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.