Dirac Delta Function - unfamiliar definition

[Orad]

Given:

f(x)=\delta(x-a)

Other than the standard definitions where f(x) equals zero everywhere except at a, where it's infinity, and that:

\int_{-\infty}^{\infty} g(x)\delta(x-a)\,dx=g(a)

Is there some kind of other definition involving exponentials, like:

\int e^{ix(k'-k)}d^3x=\delta^3(k'-k)

I remember learning something about this, but can't find a proof of it in any textbook or online at the moment, and I don't trust my memory enough to know if this is precise. Any thoughts?

[gel]

yes, the dirac delta function is proportional to the fourier transform of the constant function equal to 1. You actually get (2\pi)^3\delta^3(k).
To be rigorous, the fourier transform is defined in terms of distributions and not just standard functions. It is not a standard Riemann integral, although physicists often treat it as such.

The proof could go like the following. If f is a fourier transformable function with transform \hat f

\begin{align*}
\int f(x) \left(\int e^{ixy}\,dy\right)\,dx
&=
\int \left(\int e^{ixy}f(x)\,dx\right)\,dy\\
&= \int \hat f(y) dy
\end{align*}

Using z=0, the last line is

\int \hat f(y)e^{-iyz}\,dy = 2\pi f(z)=2\pi f(0)

-- using the inverse fourier transform. Compare this to \int f(x)\delta(x)\,dx = f(0).

This is a bit sketchy, because the integral of e^{ixy} doesn't make sense using the Riemann integral.

Also, the delta function is the derivative of the Heaviside function f(x)=1{x>0}, in the sense of distributions.

[Fredrik]

Check out this thread (http://www.physicsforums.com/showthread.php?t=242696) for a very non-rigorous argument.

[Orad]

Wow... Irrelevant, but that nicksauce guy who started the other thread is a good friend of mine in RL. I'm gonna call him now!

Ha, small world.

Oh, and I understood the (false) justifications, thanks.

[nicksauce]

Lol small world indeed