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

Let $H$ be a Hilbert space and $\sum_k x_k$ a convergent infinite sum in it. Lets say we partition the sequence $(x_k)_k$ in a sequence of blocks of finite length and change the order of summation only in those blocks, like this (for brevity illustrated only for the first two blocks) $$(x_1,\ldots,x_k,x_{k+1},\ldots,x_{k+l},\ldots )$$ becomes

$$(x_{\pi(1)},\ldots,x_{\pi(k)},x_{\gamma(k+1)},\ldots,x_{\gamma(k+l)},\ldots ),$$ where $\pi$ and $\gamma$ are permutations.

If we denote the elements of the second sequence with $x'$, does anyone know, what will happen to the series $\sum _k x'_k$ in this case, i.e. will it converge ? If yes, to what value ?

(This question is a refinement of this question, where it was shown, that the two sums are equal, if they both converge; here we know only that one of the series converges)

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

If the sizes of the blocks are bounded, then the permuted summation also converges to the same limit.

More generally, let $\sigma : \Bbb{N}_0 \to \Bbb{N}_0$ be a permutation such that $M = \sup_{n\in\Bbb{N}_0}|\sigma(n) - n|$ is finite. You can easily verify that this generalizes the former special case. Let $x'_n = x_{\sigma(n)}$ be the permuted sequence and $A_n = \{0, \cdots, n\}$. Now, from the assertion we find that

  1. $\sigma(A_n) \subset A_{n+M}$ and likewise $\sigma^{-1}(A_n) \subset A_{n+M}$, and
  2. $A_{n+M} \setminus \sigma(A_n) \subset A_{n+M}\setminus A_{n-M}$.

The first claim is trivial. To prove the second one, just observe that for $n > M$,

$$ A_{n+M}\setminus \sigma(A_n) \subset A_{n+M}\setminus \sigma(\sigma^{-1}(A_{n-M})) = \{n-M+1, \cdots, n+M\}. $$

Now we are ready. Let $(s_n)$ and $(s'_n)$ denote the partial sum of $(x_n)$ and $(x'_n)$, respectively. Then

$$\begin{align*}|s_n - s'_n| &\leq |s_{n+M} - s_n| + |s_{n+M} - s'_n| \\ &\leq \sum_{k=n+1}^{n+M} |x_k| + \sum_{k \in A_{n+M}\setminus \sigma(A_n)}|x_k| \\ &\leq \sum_{k=n+1}^{n+M} |x_k| + \sum_{k = n-M+1}^{n+M} |x_k|. \end{align*}$$

Now there are only $3M$ terms of $(x_k)$, each of which converges to 0 as $n \to \infty$. Thus by squeezing lemma, $|s_n - s'_n| \to 0$. This shows that

$$ \sum_n x'_n = \lim_n s'_n = \lim_n s_n = \sum_n x_n.$$

If $M$ is unbounded, you can easily devise a counter-example. For example, if we take the sizes of the blocks as $2, 4, 6, \cdots$ to the conditionally convergent series

$$ 1 - 1 + \frac{1}{2} - \frac{1}{2} + \frac{1}{2} - \frac{1}{2} + \frac{1}{3} - \frac{1}{3} + \frac{1}{3} - \frac{1}{3} + \frac{1}{3} - \frac{1}{3} + \cdots = 0 $$

and permute the order of the sequence in each block as

$$ 1 - 1 + \frac{1}{2} + \frac{1}{2} - \frac{1}{2} - \frac{1}{2} + \frac{1}{3} + \frac{1}{3} + \frac{1}{3} - \frac{1}{3} - \frac{1}{3} - \frac{1}{3} + \cdots, $$

it oscillates between 0 and 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.