Splitting up an infinite integral into an infinite sum of finite integrals

308 Views Asked by At

I tried splitting up an infinite integral into finite chunks, in the hope that it would help me calculate something. It didn't work, however I can't see where I am making the error. Here is an essential version of what I attempted. Is it legitimate to split up the integral like this?

I have a periodic function $f(x+nL)=f(x)$ where $n \in \mathbb{N}$. Consider the integral

$$ I=\int_{x=-\infty}^{x=+\infty} f(x) \text{d}x. $$

I split this up into chunks of length L as

$$ I = \sum_{n=-\infty}^{n=+\infty} \int_{x=nL}^{x=(n+1)L} f(x) \text{d}x. $$

Then I change the integration value to $y=x-nL$ such that

$$ I = \sum_{n=-\infty}^{n=+\infty} \int_{y=0}^{y=L} f(y) \text{d}y. $$

However now I notice that I have a divergent sum, so I think my assumption that I could split the integral up like this was wrong. Thoughts? Thank you.

1

There are 1 best solutions below

0
On

You can split the integral under the assumption that it is convergent, indeed if the integral converge than from the defenition of improper integral:

$\int_{x=-\infty}^{x=+\infty} f(x) \text{d}x =\int_{x=-\infty}^{x=0} f(x) \text{d}x + \int_{x=0}^{x=+\infty} f(x) \text{d}x$

and now,

$\int_{x=0}^{x=+\infty} f(x) \text{d}x = \lim_{n \to \infty} \int_{x=0}^{x=nL} f(x) \text{d}x = \lim_{n \to \infty} \sum_{k=0}^{n-1}\int_{x=kL}^{x=(k+1)L} f(x) \text{d}x = \sum_{k=0}^{\infty}\int_{x=kL}^{x=(k+1)L} f(x) \text{d}x$

and in the exact same way we have

$\int_{x=-\infty}^{x=0} f(x) \text{d}x = \sum_{k=-\infty}^{-1}\int_{x=kL}^{x=(k+1)L} f(x) \text{d}x$

and thus

$\int_{x=-\infty}^{x=+\infty} f(x) \text{d}x =\int_{x=-\infty}^{x=0} f(x) \text{d}x + \int_{x=0}^{x=+\infty} f(x) \text{d}x = \sum_{k=0}^{\infty}\int_{x=kL}^{x=(k+1)L} f(x) \text{d}x + \sum_{k=-\infty}^{-1}\int_{x=kL}^{x=(k+1)L} f(x) \text{d}x = \sum_{k=-\infty}^{\infty}\int_{x=kL}^{x=(k+1)L} f(x) \text{d}x$

as we want.

notice that the sum doesn't have to diverge in the case that $f$ is periodic, as the ellements of the sum can be 0. it can happen for example with $f=0$.

Just as a side note, the only periodic functions such that that $\int_{x=-\infty}^{x=+\infty} f(x) \text{d}x$ converges are functions that are almost everywhere 0.