Mathematics Stack Exchange is a question and answer site for people studying math at any level and professionals in related fields. It's 100% free, no registration required.

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

This problem is an example from Putnam and Beyond, page 118:

Let$(n_k)_{k\geq1}$ be a strictly increasing sequence of positive integers with the property that

$$\displaystyle\lim_{k \to \infty} \frac{n_k}{n_1n_2 \ldots n_{k-1}} = \infty$$

Prove that the series $\sum_{k>1}\frac{1}{n_k}$ is convergent and the its sum is an irrational number.

The proof claims that $n_{k+1} \geq 3 n_k$ for $k\geq 3$ (Edit: for $k$ sufficiently large) and then ratio test to prove the convergence of $\sum_{k>1}\frac{1}{n_k}$.

My question: If I define $n_k$ as $n_1 = 1, n_2 = 2$ and $n_{2m+1} = m\times \prod_{i=1}^{2m} n_i, n_{2m+2}=n_{2m+1}+1$ for $m \geq1$ so that $\lim\sup \frac{n_k}{n_1n_2 \ldots n_{k-1}} = \infty$ and $\lim\inf \frac{n_k}{n_1n_2 \ldots n_{k-1}} = 0$, then does it satisfy the condition?

If we use $\{n_k\}$ constructed above, I can not see why $n_{k+1} \geq 3 n_k$ for $k$ sufficiently large.

Edit: My $\epsilon-\delta$ interpretation of $\lim a_k = \infty$ is that $\forall N \in \mathbb{R}^{+} \text{ and } \forall n \in \mathbb{N}, \exists n_0 \ge n $ such that $a_{n_0} \geq N$.

share|cite|improve this question
In your last statement, the $\forall n\in\mathbb N$ isn't meaningful - there isn't an $n$ in the rest of the statement. – Greg Martin Apr 10 '13 at 16:33
@GregMartin: typo fixed – Lei Lei Apr 10 '13 at 18:19

With your definition, it's clearly true that $\liminf \frac{n_k}{n_1n_2\cdots n_{k-1}} \le 1$. But that doesn't seem relevant to the stated problem, given its hypothesis.

It's not necessarily true that $n_{k+1} \ge 3n_k$ for $k\ge3$ (the sequence can start with $n_k=k$ for as many terms in a row as you like), but there does exist $k_0$ such that it's true for $k\ge k_0$.

share|cite|improve this answer
probably the proof says that without loss of generality you can ignore the initial terms that don't respect the condition – suissidle Apr 10 '13 at 6:27
So $n_k$ from my example does not satisfies the hypothesis of the stated problem? – Lei Lei Apr 10 '13 at 7:23
Definitely not. – Greg Martin Apr 10 '13 at 16:34
@GregMartin: I guess I know why. I interpret $\lim a_k = \infty$ as $\{a_k\}_{k \geq 1}$ as a divergent sequence, not as a convergent sequence in extended real number system. – Lei Lei Apr 10 '13 at 18:18
Good diagnosis: the symbols $\lim a_k=\infty$ indicates a sequence that diverges in a very specific way. – Greg Martin Apr 10 '13 at 18:28

let $\varepsilon=3$ then with the definition of limit there must be a $n_0\in N$ such that for any $n>n_0$ there has $n_{k+1}>3n_kn_{k-1}n_{k-2}...n_2n_1>3n_k$

and $n_0 = 3$ has no effects on how solving this problem

share|cite|improve this answer
that is the definition for convergent sequence. I think for divergent sequence, it is $\forall n \exists n_0 > n$ such that $n_{k+1} > 3n_k n_{k-1} \ldots n_2 n_1$. – Lei Lei Apr 10 '13 at 7:39
you can consider infinity as a number .but in fact we say a same thing – Xiaolang Apr 10 '13 at 8:00
your definition fails for the example I constructed, but my definition still works. – Lei Lei Apr 10 '13 at 16:03

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

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