Root mean square (RMS) of a sinc function

AI Thread Summary
The discussion focuses on calculating the root mean square (RMS) of the sinc function, noting that the RMS value approaches zero due to the function's aperiodicity and its energy being spread over an infinite interval. The RMS definition is provided, and the integral for the sinc function is evaluated, leading to the conclusion that the RMS remains zero as "a" approaches infinity. Participants express skepticism about the relevance of RMS for aperiodic functions and discuss potential approximations for the sinc function, suggesting that while one approximation is 1/(pi r), it lacks utility as it is not square-integrable. The conversation highlights the challenges in applying RMS to functions like sinc that do not have finite periods.
redtree
Messages
335
Reaction score
15
Given that the root mean square (RMS) of a sine function is as follows:

RMS of (a*sin(\omega*r) = a / \sqrt{}2

Let a = 1/\omega

Thus

RMS of ((1/\omega)*sin(\omega*r)) = 1 / (\omega*\sqrt{}2)

But for sinc(\omega*x), what is formula for the RMS?
 
Mathematics news on Phys.org
The rms value of sinc is zero.

I'll define rms as
f_{rms} = \sqrt{ \lim_{a \to \infty} \frac{1}{2a} \int_{-a}^{a} f(x)^2 \,dx }.
Let's find the rms for sinc(x). The integral here is
\int_{-a}^{a} \frac{\sin^2 x}{x^2} \,dx
= \int_{-a}^{-1} \frac{\sin^2 x}{x^2} \,dx + \int_{-1}^{1} \frac{\sin^2 x}{x^2} \,dx + \int_{1}^{a} \frac{\sin^2 x}{x^2} \,dx
\le \int_{-a}^{-1} \frac{1}{x^2} \,dx + \int_{-1}^{1} 1 \,dx + \int_{1}^{a} \frac{1}{x^2} \,dx
= 4 - \frac2a < 4.

Thus, the rms value of sinc is
\sqrt{ \lim_{a \to \infty} \frac{1}{2a} \int_{-a}^{a} \frac{\sin^2 x}{x^2} \,dx } \le \sqrt{ \lim_{a \to \infty} \frac{1}{2a} 4 } = 0.
 
Thanks
 
I understand the case when a approaches infinity, but what about the case where "a" is a finite number?
 
redtree said:
I understand the case when a approaches infinity, but what about the case where "a" is a finite number?

The sinc function is not periodic, in other words it's period is infinity. That's why adriank found it's RMS value to be zero. Think of it this way, it's got a finite total energy that is spread over an infinite time interval, the result is an RMS value of zero.

In many ways it doesn't really make a lot of sense to even talk about the RMS value of an aperiodic function like that. What exactly is your application?
 
I want to find a good approximation for the following function:
a = k * sinc( pi k r) = sin (pi k r) / (pi r)

One approximation would be the following
a = 1 / (pi r)

But I think a better approximation is based on RMS. My feeling is that the sinc funciton above is a periodic function, though with decreasing amplitude as a function of r and therefore can be approximated using RMS.
 
Well, the rms value of sin is 1/\sqrt{2}, so you might guess
a = \frac{1}{\sqrt{2} \lvert \pi r \rvert}.
It doesn't seem very useful though. It's not square-integrable, for instance.
 
Back
Top