Mathematics Stack Exchange is a question and answer site for people studying math at any level and professionals in related fields. Join them; it only takes a minute:

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 is I think a very simple question about infinite sequences. I thought I knew the answer but the manipulation described below worries me.

Suppose I divide the interval $(0,\frac{1}{4})$ into infinitely many subintervals $S_n = (\frac{1}{(n+1)^2},\frac{1}{(n)^2})$, to wit: $S_1 = (\frac{1}{9},\frac{1}{4}),S_2 = (\frac{1}{16},\frac{1}{9})$, etc. Suppose there is a countably infinite subsequence $\{S_{n_i} \}$ of these intervals that interests me. I wish to segregate this subsequence by moving it to the left of the interval, so that the two countable sequences $\{S_n\} \setminus \{S_{n_i}\}$ and $\{S_{n_i}\}$ are segregated and remain in length-order, respectively. So we would have, $0,...S_{n_2},S_{n_1},0,...,S_2,S_1$.

Are any special assumptions needed to justify this manipulation (and is the situation clear)?

Thanks for any help.

Edit, example:

We have the line segment s: 0 _______ 1/4

I divide it into subintervals as described. Now suppose I want to take the subset of intervals indexed by odd n, and move them to the left of the segment. From the right at x = 1/4, I have a subsequence of intervals whose length approaches zero near (let us say) x = s, and then a new subsequence that begins at s, whose lengths approach 0 as they move towards x = 0. Does this help?

share|cite|improve this question
up vote 1 down vote accepted

Revised: Let $a_n$ be the length of $S_n$, so that $\sum_{n\ge 1}a_n=\frac14$. This is an absolutely convergent series, so we may rearrange the terms as we wish. Let $\langle n_k:k\in\Bbb Z^+\rangle$ be an increasing enumeration of the subscripts of the intervals that are to be ‘moved’ to the left, and let $\langle m_k:k\in\Bbb Z^+\rangle$ be an increasing enumeration of the subscripts of the remaining intervals.

For $k\in\Bbb Z^+$ let $s_k=\sum_{i=1}^ka_{m_i}$; $s_k$ is the total length of the first $k$ of the remaining intervals. Let $x_0=\frac14$, and for $k\in\Bbb Z^+$ let $x_k=\frac14-s_k$ and $T_k=(x_k,x_{k-1})$. Let $s=\sum_{i\ge 1}a_{m_i}$; then $0<s<\frac14$, and

$$\bigcup_{k\ge 1}T_k=\left(s,\frac14\right)\setminus\{x_k:k\in\Bbb Z^+\}\;.$$

Now repeat the process with the $S_{n_k}$ in the interval $(0,s)$. For $k\in\Bbb Z^+$ let $u_k=\sum_{i=1}^ka_{n_k}$. Let $y_0=s$, and for $k\in\Bbb Z^+$ let $y_k=s-u_k$ and $U_k=(y_k,y_{k-1})$. Then $\sum_{k\ge 1}a_{n_k}=\frac14-s$, since $\sum_{k\ge 1}a_k=\frac14$, and

$$\bigcup_{k\ge 1}U_k=(0,s)\setminus\{y_k:k\in\Bbb Z^+\}\;.$$

The intervals $U_k$ and $T_k$ are now arranged as you want them.

share|cite|improve this answer
@daniel: Yes, I now understand what you’re trying to do; the problem was that your original description didn’t distinguish between the intervals and their lengths. See my revised answer. – Brian M. Scott Jul 14 '12 at 7:41
Yes, this answers my question, much appreciated. – daniel Jul 14 '12 at 7:43

Introduce $x_i=\sum\limits_{j=i}^{+\infty}\left(\frac1{n_j^2}-\frac1{(n_j+1)^2}\right)$. Then, for every $i\geqslant1$, the interval $S_{n_i}=\left(\frac1{(n_i+1)^2},\frac1{n_i^2}\right)$ gets moved to the interval $X_i=(x_{i+1},x_i)$.

To get the new location of each interval $S_n$ with $n$ not in the sequence $(n_i)_{i\geqslant1}$, write down the collection of integers not in the sequence $(n_i)_{i\geqslant1}$ as an increasing sequence $(m_i)_{i\geqslant1}$ and introduce $y_i=\sum\limits_{j=i}^{+\infty}\left(\frac1{m_j^2}-\frac1{(m_j+1)^2}\right)$. Then, for every $i\geqslant1$, the interval $S_{m_i}=\left(\frac1{(m_i+1)^2},\frac1{m_i^2}\right)$ gets moved to the interval $Y_i=(x_1+y_{i+1},x_1+y_i)$.

One can check that $X_{i+1}\lt X_i\lt Y_{j+1}\lt Y_j$ for every $i\geqslant1$ and $j\geqslant1$, with the leftmost point of $X_i$ going to $0$ when $i\to\infty$, the rightmost point of $X_1$ being $x_1$, the leftmost point of $Y_j$ going to $x_1$ when $j\to\infty$, and the rightmost point of $Y_1$ being $x_1+y_1=1$.

share|cite|improve this answer

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.