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 was wondering what is the justification for this step(changing the indexes)$\displaystyle\sum_{n=0}^{\infty}\frac{a^{n}}{n!}\sum_{m=0}^{\infty}\frac{b^{m}}{m!}=\sum_{k=0}^{\infty}\frac{1}{k!}\sum_{n=0}^{k}\frac{k!}{n!(k-n)!}a^{n}b^{k-n}$ , in Rudin's Real and complex analysis prolog to show $(\exp{a})( \exp{b})=\exp{(a+b)}$, is it the same principle that use fubini's theorem for integrals, I mean that one that says if given the domain of integration D=AxB=ExF then I can do something like $\int_{D}=\int_{A}\int_{B}=\int_{E}\int_{F}$ , I would appreciate any hint or reference to this, thanks in advance.

share|cite|improve this question
Yes you are right. It can be seen as Fubini's theorem for series. – Ragib Zaman Sep 28 '11 at 5:32
@RagibZaman but I got a problem seeing if the "domain" of the indexes in the RHS is the same domain in the LHS of the equation. LHS is like all the plane $R^{2}$ and RHS is like the area down the identity function. Am I right? – Ivan3.14 Sep 28 '11 at 5:43

If we let $f_n = \frac{a_n}{n!}$ and $g_m = \frac{b_m}{m!}$, you're asking why $$\displaystyle\sum_{n=0}^{\infty} f_n \sum_{m=0}^{\infty} g_m = \sum_{k=0}^{\infty}\frac{1}{k!}\sum_{n=0}^{k} k! f_n g_{k-n} = \sum_{k=0}^{\infty}\sum_{n=0}^{k}f_n g_{k-n}$$

The reason is that the expression $$\sum_{n=0}^{\infty} f_n \sum_{m=0}^{\infty} g_m $$ is adding up the following numbers row-by-row,

$$ \begin{matrix} f_0 g_0 & f_0 g_1 & f_0 g_2 & f_0 g_3 & \cdots \\ f_1 g_0 & f_1 g_1 & f_1 g_2 & f_1 g_3 & \cdots \\ f_2 g_0 & f_2 g_1 & f_2 g_2 & f_2 g_3 & \cdots \\ f_3 g_0 & f_3 g_1 & f_3 g_2 & f_3 g_3 & \cdots \\ \vdots & \vdots & \vdots & \vdots & \ddots \end{matrix}, $$ while the sum $$\sum_{k=0}^{\infty}\sum_{n=0}^{k}f_n g_{k-n}$$ is adding up the same numbers diagonal-by-diagonal.

Added: The answer above intentionally ignores questions of convergence. It assumes that the OP's question must be about the algebra of why the equality is true, since if there were problems with convergence the equality wouldn't be asserted in Rudin. However, for those concerned about convergence issues here, Merten's theorem can be applied, as wnoise points out below.

share|cite|improve this answer
And there are no issues with convergence... – wnoise Sep 28 '11 at 6:56
@wnoise : Why there is no issue with convergence? What justifies that?\ – Arjang Sep 28 '11 at 7:34
@Arjang I Think wnoise was being sarcastic. – Ragib Zaman Sep 28 '11 at 7:47
@RagibZaman : LOL, I was staring at it trying to remember the double series convergence criterion! – Arjang Sep 28 '11 at 7:57
I was apparently too elliptical. I meant that this answer is true in the sense that this is the algebra for why we do this, but like any rearrangement of infinite sums it requires that there are no issues with convergence. In this particular case, we have the product of two absolutely convergent series, and Merten's theorem gives convergence for the infinite sum on k of the finite sum on n. I think we can actually show absolute convergence by showing that the magnitude of terms goes to zero quickly enough as $n+m=k$ increases. – wnoise Sep 28 '11 at 9:05

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.