Re: Gaussian distribution
- From: "Randy Poe" <poespam-trap@xxxxxxxxx>
- Date: 22 Mar 2007 15:07:46 -0700
On Mar 22, 5:34 pm, Kaba <REkalleMOunderscoreVErutane...@xxxxxxxxxxx>
wrote:
Hello
How do you prove that a Gaussian distribution integrates to 1 over -oo..
+oo?
That aside, I'm trying to find the fourier transform of a Gaussian
distribution, how do you do that?
My problems lie in the term of the form e^(bx^2) which I can't seem to
be able to integrate. I tried a change of variables x(u) = sqrt(u),
didn't take me anywhere. I also tried integration by parts two times
sequentially, didn't also take me anywhere.
This is not homework. I'd like to know the frequency response of a
Gaussian filter so I can design them using a desired cutoff frequency.
There's a trick to it.
Let I = integral(-inf, inf) exp(-x^2) dx
Multiply I by itself.
I^2 = [integral(-inf, inf) exp(-x^2) dx]*
[integral(-inf, inf) exp(-y^2) dy]
= integral(-inf,inf) integral(-inf,inf) exp[-(x^2+y^2)] dx dy
Now the double integral over dx dy can be changed to polar
coordinates. The exponential becomes exp(-r^2), but
because dx dy becomes r dr d(theta), a change of
variables to u=r^2 gives you an easy integral of
exp(-u) du
- Randy
.
- References:
- Gaussian distribution
- From: Kaba
- Gaussian distribution
- Prev by Date: Cantor meets Gödel at the village barber's
- Next by Date: online programming course
- Previous by thread: Re: Gaussian distribution
- Next by thread: Re: Gaussian distribution
- Index(es):
Relevant Pages
|