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

I have a problem I do not find a solution. Given two series $\left(a_n\right)_{n \in \mathbf{Z}}$ and $\left(b_n\right)_{n \in \mathbf{Z}}$ which have a cauchy product $\left(c_n\right)_{n \in \mathbf{Z}} = \left(\sum_{i \in \mathbf{Z}}{a_i b_{n-i}}\right)_{n \in \mathbf{Z}}$ and $\forall n,\sum_{i\in\mathbf{Z}}{\left|a_ib_{n-i}\right|}<\infty$.

Do $\left(c_n\right)_{n \in \mathbf{Z}}=0$ implies $\left(a_n\right)_{n \in \mathbf{Z}}=0$ or $\left(b_n\right)_{n \in \mathbf{Z}}=0$?

It is true if $\exists n_0, \forall n<n_0, a_n=0$ (like in traditional formal Laurent series). I can not find a demonstration or a counterexample for the general case.

share|cite|improve this question
Your sum over $i\in \mathbb Z$ need not be well defined without further restrictions - in what order are the terms to be added? – Mark Bennet May 31 '13 at 20:16
I supposed implicitely that $\forall n, \sum_{i\in\mathbf{Z}}{\left|a_ib_{n-i}\right|}$ is defined. I add it to the question. – oao May 31 '13 at 20:20
What if $a_n=1$ for all $n$ and $b_0=1, b_1=-1$ and $b_n=0$ otherwise? – Hagen von Eitzen May 31 '13 at 20:25
It is a good idea, but $\sum{\left|a_i b_{n-i}\right|}$ does not converge. – oao May 31 '13 at 20:29

Just look at some Fourier series (two functions with product zero), and use their coefficients. Example: $$ a_0 = \frac{1}{2},\qquad a_{n} = \frac{1}{\pi i n}, n \text{ odd}, \qquad a_n = 0, \text{otherwise} $$

$$ b_0 = \frac{1}{2},\qquad b_{n} = \frac{-1}{\pi i n}, n \text{ odd}, \qquad b_n = 0, \text{otherwise} $$

share|cite|improve this answer
I am not sure to understand. $c_0 = \sum_j{a_j b_{-j}} = 1/4 + \sum_{j \neq 0}{\frac{1}{-(2j+1)^2}\frac{1}{-\pi}} = 0$, but $c_1 = 0 + \sum_{i \neq -1,1}{\frac{1}{(i+1)(i-1)}\frac{1}{-\pi}}\neq 0$. – oao May 31 '13 at 22:30
The last term is of course c_1 = 0 + \sum_{i \neq -1,1}{\frac{1}{(2i+1)(2i-1)}\frac{1}{-\pi}}\neq 0. Too late to edit. – oao May 31 '13 at 22:40
The sum of any two odd number is even, but $1$ is odd. So $c_1 = a_0b_1+a_1b_0$. – GEdgar Jun 1 '13 at 12:14

Proof by induction:

$n=0$. If $a_0b_0=0$, one of them is zero. If both are zero, repeat the problem with sums from $1$ to $\infty$ until the first non-zero one is found. If all are zero, we are done.

We assume $a_0 = 0$ and $b_0 \ne 0$.

$n > 0$. Suppose $c_n = \sum_{i=0}^n a_i b_{n-i} = 0$ and $a_i = 0$ for $i$ from $0$ to $n-1$. Then $a_i b_{n-i} = 0$ for $i$ from $0$ to $n-1$, so $0 = c_n = a_n b_0$. Since $b_0 \ne 0$, $a_n = 0$.

Therefore all the $a_n = 0$.

share|cite|improve this answer
It is not possible to do that, the sum $c_n=\sum_{i \in\mathbf Z}{a_i b_{n-1}}$ are on an infinite number of items. – oao May 31 '13 at 22:53
This is the solution for index set $\mathbb N$. And (essentially) what the OP means by "traditional formal Laurent series". – GEdgar Jun 1 '13 at 14:40

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.