1
$\begingroup$

Show that a conditionally convergent series has a rearrangement converging to $+\infty$


Thoughts:

  1. A conditionally convergent series is a series that converges but not absolutely converges $\lim_{m\to\infty} \sum_{n=0}^{m} a_n$ exists, $\sum_{n=0}^{\infty} |a_n|=\infty$

  2. if $\sum_{n=0}^{\infty} a_n$ is a conditionally convergent series, then for every real number $L$, there is a rearrangement that converges to $L$

Since we are given $\sum_{n=0}^{\infty} a_n$ converges conditionally. Intuitively, we obtain one positive and one negative series, which should be converge to the same limit.

$\endgroup$
7
  • 1
    $\begingroup$ The last statement is somewhat vague.. The positive and negative series diverge. Unless you mean to say one converge to $\infty$ and the other to $-\infty$. $\endgroup$
    – Maesumi
    Feb 27 '13 at 23:28
  • 2
    $\begingroup$ Do you know how to make a conditionally convergent series sum to a given $L\in\mathbb{R}$? $\endgroup$
    – Stahl
    Feb 27 '13 at 23:29
  • 2
    $\begingroup$ Suppose your series, when separated into positive and negative parts is formally $\sum p_i$+$\sum n_j$. Add enough $p_i$'s until you exceed $1$ then add just $n_1$. Then add enough $p_i$'s until you exceed $2$ and add $n_2$ and so on add enough positive ones until you exceed $k$ and then add $n_k$. The resulting series diverges to $+\infty$ and has all original terms in it. $\endgroup$
    – Maesumi
    Feb 27 '13 at 23:34
  • 1
    $\begingroup$ I suppose @Paul wants a formal proof. $\endgroup$
    – Maesumi
    Feb 27 '13 at 23:40
  • 1
    $\begingroup$ Suppose $L\in\mathbb{R}^+$. The way we normally construct the rearrangement is by arranging the positive terms $p_i$ and negative terms $n_i$ so that $\left|p_i\right|\geq\left|p_{i+1}\right|$, and $\left|n_i\right|\geq\left|n_{i+1}\right|$. Then we add $p_1 + p_2 + \ldots + p_n$ such that $n$ is the smallest integer with $p_1 + p_2 + \ldots + p_n \geq L$. Then add $n_1$. Proceed to add more $p_i$'s until the sum again exceeds $L$, and repeat. This will give you a series that sums to $L$. @Maesumi's argument is an adaption to the case $L = \infty$. $\endgroup$
    – Stahl
    Feb 27 '13 at 23:47
1
$\begingroup$

This is a particular case of the astonishing Riemann Series Theorem . Note that you can rearrange a conditional convergent series in such a way as to make the rearranged series converge to whatever you want.

$\endgroup$
2
  • $\begingroup$ would you explain a little bit more? $\endgroup$
    – Paul
    Feb 28 '13 at 1:40
  • 2
    $\begingroup$ Read the theorem in the link: if $\,\sum a_n\,$ is a conditionally convergent series, then for any $\,\alpha\in\Bbb R\cup\{\pm\infty\}\,$ there exists a permutation $\,\sigma\in S_{\Bbb N}\,$ s.t. $\,\sum a_{\sigma{n}}=\alpha\,$ ...This means that upon being given any real number or $\,\pm\infty\,$ , you can arrange the series's terms as to get a new series that converges to that...amazing, uh?! $\endgroup$
    – DonAntonio
    Feb 28 '13 at 2:20
1
$\begingroup$

We have that $\sum_n \frac{(-1)^n}{n}-\frac{(-1)^n}{n} = 0$, but $\sum_n \frac{2}{n} = \infty$. Then $(c_n)$ where

\begin{align} c_0 &= 1, \\ c_1 &= -1, \\ f(n) &= \max_{k < n} \left\{\frac{1}{c_k}\right\}+1, \\ g(n) &= \min_{k < n} \left\{\frac{1}{c_k}\right\}-1, \\ c_n &= \begin{cases} \frac{1}{f(n)} & \text{if } \sum_{k<n} c_k < -g(n)\\ \frac{1}{g(n)} & \text{if } \sum_{k<n} c_k \geq -g(n) \end{cases} \end{align}

would be your desired sequence. Have fun ;-)

$\endgroup$

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.