Proving the scaling property of the Delta function

Join the discussion
Ask a follow-up here, or get your own question answered by working scientists, mathematicians and engineers — people, not an autocomplete.
Real named experts · corrections over time · the nuance an AI answer skips
3 replies · 25K views
caesius
Messages
22
Reaction score
0

Homework Statement



Prove that [tex]\delta(at)=\frac{1}{abs(a)}\delta(t)[/tex]

Hint: Show that [tex]\int\phi(t)\delta(at)dt=\frac{1}{abs(a)}\phi(0)[/tex]

(the limits of integration are from -inf to +inf btw, I couldn't find how to put them in..)

Homework Equations





The Attempt at a Solution



Ok. I understand that the integral is only defined for at = 0, i.e., t = 0. But if I follow this logical then I evaluate the integral to be phi(0), not phi(0)/a. Where does the a come from?

And I'm not entirely sure how proving that relationship helps with the proof of the scaling function...
 
Physics news on Phys.org
That trick only works when the argument to the delta is the bare integration variable.
So try substituting u = a t, so you get something like
[tex]\int_{-\infty}^{\infty} f(u) \delta(u) \, du = f(0)[/tex]
(and click the formula to see how I did the boundaries)
 
Ok I've substituted x = at, now I get

[tex] \int_{-\infty}^{\infty}\phi(\frac{x}{a})\delta(x)d\frac{x}{a} = \frac{1}{a}\int_{-\infty}^{\infty}\phi(\frac{x}{a})\delta(x)dx = \frac{1}{a}\phi(\frac{0}{a}) = \frac{1}{a}\phi(0)[/tex]

I don't quite see how I'm allowed to take the 1/a out of the d/dx operater and put it at the front though. And why should it be the abs(a) not a?

And I still can't see how this helps prove anything.

Cheers, Benjamin
 
You are allowed to take the a outside, because it is just a constant: du = a dt, so dt = (1/a) du, and you can put constants outside the integration as in
[tex]\int \tfrac12 x^2 \, dx = \tfrac12 \int x^2 \, dx[/tex]

The absolute value of a comes from transforming the integration boundaries. If a < 0, then -infinity goes to +infinity in the transformed integral, and vice versa. However, in the final formula you simply want -infinity to be the lower boundary, and +infinity as the upper boundary. So if a < 0 you have to swap them, at the cost of a minus sign, whereas for a > 0 you don't. Do you have to multiply the whole thing by sgn(a), and then you can shorten sgn(a) / a to 1/abs(a).