3
$\begingroup$

I have been told that, if $\{a_n\}_{n\in\mathbb{N}}$, $\{a_{-n}\}_{n\in\mathbb{N}^+}$, $\{b_n\}_{n\in\mathbb{N}}$ and $\{b_{-n}\}_{n\in\mathbb{N}^+}$ are absolutely summable complex sequences, then$$\Bigg(\sum_{n=-\infty}^{\infty}a_n\Bigg)\Bigg(\sum_{n=-\infty}^{\infty}b_n\Bigg)=\sum_{n=-\infty}^{\infty}\sum_{k=-\infty}^{\infty}a_{n-k}b_k$$where $\sum_{n=-\infty}^{\infty}a_n=\sum_{n=0}^{\infty}a_n+\sum_{n=1}^{\infty}a_{-n}$.

I know that if $\{a_n\}_{n\in\mathbb{N}}$ or $\{b_n\}_{n\in\mathbb{N}}$ is absolutely summable, then $(\sum_{n=0}^{\infty}a_n)(\sum_{n=0}^{\infty}b_n)=\sum_{n=0}^{\infty}\sum_{k=0}^{n}a_{n-k}b_k=\sum_{n=0}^{\infty}\sum_{k=0}^{\infty}a_{n-k}b_k$, i.e. the proposition is true if $\forall n\leq-1\quad a_n=0=b_n$, and have tried to use that to prove the general case, but I get nothing. I have not studied any theory of measure yet.

I $+\infty$-ly thank you for any help!!!

EDIT: my question had been considered as a duplicate to a question already asked, but that is about the case $(\sum_{n=0}^{\infty}a_n)(\sum_{n=0}^{\infty}b_n)=\sum_{n=0}^{\infty}\sum_{k=0}^{n}a_{n-k}b_k$, which I do know, as I had written. Though, I cannot generalise that to show that $(\sum_{n=-\infty}^{\infty}a_n)(\sum_{n=-\infty}^{\infty}b_n)$, which I hope I correctly write as $$(\sum_{k=0}^{\infty}a_k+\sum_{k=1}^{\infty}a_{-k}) (\sum_{k=0}^{\infty}b_k+\sum_{k=1}^{\infty}b_{-k})$$ $$=\sum_{n=0}^{\infty}\sum_{k=0}^{n} a_{n-k}b_k +\sum_{n=0}^{\infty}\sum_{k=0}^{n} a_{n-k}b_{-k-1}+\sum_{n=0}^{\infty}\sum_{k=0}^{n} a_{-n+k-1}b_{k}+\sum_{n=1}^{\infty}\sum_{k=1}^{n} a_{-n+k-1}b_{-k},$$

is equal to $\sum_{n=-\infty}^{\infty}\sum_{k=-\infty}^{\infty}a_{n-k}b_k$. The people who regarded this question of mine as a duplicate to that were enough to have this question closed for a while, therefore, I suppose that it may well be trivial that the equality $(\sum_{n=-\infty}^{\infty}a_n)(\sum_{n=-\infty}^{\infty}b_n)=\sum_{n=-\infty}^{\infty}\sum_{k=-\infty}^{\infty}a_{n-k}b_k$ derives from the equality, known by me, $(\sum_{n=0}^{\infty}a_n)(\sum_{n=0}^{\infty}b_n)=\sum_{n=0}^{\infty}\sum_{k=0}^{n}a_{n-k}b_k$. If that is trivial, I do not realise that fact: could anybody, either from those considering the question identical to that or anybody else, please, show me how to derive the equality with the indices $n$ from $-\infty$ to $+\infty$? I... $+\infty$-ly thank you! ;-)

$\endgroup$
  • 1
    $\begingroup$ This is essentially Fubini's theorem on the measure space $\Bbb Z^2$ with the counting measure. The proof of this case will differ from the proof in the question you linked to only in the bookkeeping of indices. $\endgroup$ – Greg Martin Sep 18 '14 at 16:11
  • 1
    $\begingroup$ @GregMartin If both are summable [absolutely convergent, here], then it is indeed straightforward. If one of the series is only conditionally convergent, it is not so straightforward (and I'm not even 100% sure it works then). $\endgroup$ – Daniel Fischer Sep 18 '14 at 16:27
  • $\begingroup$ Thank you both!!! I've edited to cover the both absolutely convergent case only. I have tried to apply the proof of linked question to this case, but I have had serious difficulties, since we have four series to handle here: $U=u_0,+u_1+u_2+...$ $+u_{-1},+u_{-2}+...$, $V=v_0,+v_1+v_2+...$ $+v_{-1},+v_{-2}+...$, whose product I wish to prove to be... $\endgroup$ – Self-teaching worker Sep 18 '14 at 19:26
  • $\begingroup$ $UV=\sum_{n=0}^{+\infty}(\sum_{k=0}^{+\infty}u_{n-k}v_k+\sum_{k=-\infty}^{-1}u_{n-k}v_k)+\sum_{n=-\infty}^{-1}(\sum_{k=0}^{+\infty}u_{n-k}v_k+\sum_{k=-\infty}^{-1}u_{n-k}v_k)=(u_0v_0+u_{-1}v_1+u_{-2}v_2+...+u_1v_0+u_0v_1++u_{-1}v_{2}+...)+(u_1v_{-1}+u_2v_{-2}+u_3v_{-3}+...+u_2v_0+u_3v_{-1}+u_4v_{-2}+...)+(u_{-1}v_0+u_{-2}v_1+u_{-3}v_2+...u_{-2}v_0+u_{-3}v_1+u_{-4}v_2+...)+(u_0v_{-1}+u_1v_{-2}+u_2v_{-3}+...+u_{-1}v_{-1}+u_0v_{-2}+u_1v_{-3}+...)$. $\endgroup$ – Self-teaching worker Sep 18 '14 at 19:38
  • 2
    $\begingroup$ You really ought to reformat this question - it is really hard to read. Too many complex $\sum$ operations in the middle of paragraphs. $\endgroup$ – Thomas Andrews Sep 19 '14 at 18:12
1
+50
$\begingroup$

Because the series are absolutely convergent, then so are $a(z)=\sum_{n=-\infty}^{\infty}a_{n}z^{n}$ and $b(z)=\sum_{n=-\infty}^{\infty}b_{n}z^{n}$ absolutely convergent for $|z|=1$. Therefore, the following is also absolutely convergent for $|z|=1$: $$ a(z)b(z) = \sum_{n=-\infty}^{\infty}\sum_{m=-\infty}^{\infty}a_{n}z^{n}b_{m}z^{m}. $$ So it is permissible to arbitrarily rearrange the terms of the series, without changing the value of the sum. In particular, one may collect all like powers of $z$ without changing the value of the sum. The coefficient of $z^{k}$ is $\sum_{n=-\infty}^{\infty}a_{n}b_{k-n}$ because $n+(k-n)=k$. Therefore, $$ a(z)b(z) = \sum_{k=-\infty}^{\infty}\left[\sum_{n=-\infty}^{\infty}a_{n}b_{k-n}\right]z^{k}. $$ The powers of $z$ were introduced only to illustrate how to collect the terms, and to know that none of the terms were omitted from the final sum. Finally, set $z=1$.

$\endgroup$
  • $\begingroup$ Considering the multiple non-generalised (from $k=0$ or $k=1$ to $\infty$) series confused me. Enlightening answer. Thank you so much! $\endgroup$ – Self-teaching worker Sep 22 '14 at 9:17
  • 1
    $\begingroup$ @DavideZena : You're very welcome. $\endgroup$ – DisintegratingByParts Sep 22 '14 at 11:49
3
$\begingroup$

This is about $L^1({\mathbb Z}^2,\#)$.

When $S$ is a countable set ($S={\mathbb Z}^2$ in our example) a function $f:\ S\to{\mathbb C}$ is summable when $$\sup_{J\subset S,\ J\ {\rm finite}}\sum_{x\in J} |f(x)|<\infty\ .$$ If this is the case then it follows from the completeness of ${\mathbb C}$ that there is a unique $s\in{\mathbb C}$ with the following property: For each $\epsilon>0$ there is a finite set $J_0\subset S$ such that for all finite sets $J$ with $J_0\subset J\subset S$ one has $$\left|\sum_{x\in J}f(x)-s\right|<\epsilon\ .$$ Denote this number $s$ by $\int_Sf$.

Given two absolutely convergent complex series $$\sum_{k\in{\mathbb Z}}a_k=:A, \quad \sum_{k\in{\mathbb Z}}b_k=:B$$ it is easy to check that $$f:\ {\mathbb Z}^2\to{\mathbb C},\qquad(j,k)\mapsto a_j\>b_k$$ is summable and that $$\int_{{\mathbb Z}^2}f=A\cdot B\ .$$ On the other hand, for summable $f$ we have a Fubini theorem: Assume that $f$ is summable over $S$, and that $S=\bigcup_{\iota\in I} S_\iota$ is any (maybe infinite) partition of $S$. Then $$\int_S f=\sum_{\iota\in I} s(\iota)\ ,$$ where $$s(\iota)=\int_{S_\iota} f\ .$$ Now apply this to the partition $${\mathbb Z}^2=\bigcup_{n=-\infty}^\infty\bigl\{(n-k,k)\>|\> k\in{\mathbb Z}\bigr\}\ .$$

$\endgroup$
  • $\begingroup$ I heartily thank you, dear Professor! A constructive and overwhelmingly interesting answer. I don't know anything about the theory of measure until now, though I plan to study it on Kolmogorov-Fomin's. Could the equality be proven with more elementary methods? As to the fact that "it follows from the completeness of ${\mathbb C}$ that there is a unique $s\in{\mathbb C}$ such that...", how do you know that (I do know some topology and theory of metric spaces)? $\infty$ thanks again!!! $\endgroup$ – Self-teaching worker Sep 20 '14 at 20:17

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.