On the theory of moving series inside integrals

37 Views Asked by At

I'm working on my bachelor's thesis on the Gibbs phenomenon and in one of my sources the following step is made during a proof of the existence of the Gibbs phenomenon for the ramp function: $$ \sum_{n=1}^{\infty}\int_{0}^{x}\cos(nt)\,\mathrm{d}t = \lim_{n\to\infty} \int_{0}^{x} \left( \frac{ 1 }{ 2 } + \sum_{k=1}^{n}\cos(kt)\right)\,\mathrm{d}t - \frac{x}{2}\,. $$ For some reason, I can't for the life of me figure out what the author did here. Obviously the series was moved inside of the integral, but where do the $\frac{1}{2}$ and $-\frac{x}{2}$ come from? Was a specific trig identity used here or what?

Any help or suggestions on possible sources to consult would be much appreciated.

1

There are 1 best solutions below

2
On BEST ANSWER

The $\frac12$ and $\frac x2$ cancel, so I think that's a separate step to what the author did. It's a standard trick to add something, then subtract the same thing to make other simplifications possible down the line. My guess is that this is the reason for this manouver: either the $\frac12$ or the $\frac x2$ is going to be useful to simplify something else a few lines down.

As to the sum and integral, $\sum_{n = 1}^\infty$ by definition means $\lim_{k\to \infty}\sum_{n = 1}^k$ (and then for some reason he swapped $n$ and $k$ around). In any of these finite sums, you can without problem swap the order of $\sum$ and $\int$. What you can't do is change the order of $\int$ and $\lim$.