Re: How to get this gaussian integral result?
- To: mathgroup at smc.vnet.net
- Subject: [mg125820] Re: How to get this gaussian integral result?
- From: Scott Hemphill <hemphill at hemphills.net>
- Date: Wed, 4 Apr 2012 04:28:08 -0400 (EDT)
- Delivered-to: l-mathgroup@mail-archive0.wolfram.com
- References: <jledpe$h35$1@smc.vnet.net>
- Reply-to: hemphill at alumni.caltech.edu
simplerbysimpler at gmail.com writes:
> Integrate[
> x1^(n1) x2^(n2) x3^(n3) Exp[
> a x1 x1 + b x2 x2 + c x3 x3 + d12 x1 x2 + d23 x2 x3 + d13 x1 x3 +
> d], {x1, -Infinity, Infinity}, {x2, -Infinity,
> Infinity}, {x3, -Infinity, Infinity}]. n1,n2,n3 are Natural numbers. I still cannot get the general result ,although mathematica takes so much time. Can you help me about this?
I think the basic problem is that this integral doesn't converge for
most of the values of a, b, c, d12, d23 and d13. If I wanted the value
when it converges, I would change variables in such a way that the "Exp"
expression reads "Exp[- u1 u1 - u2 u2 - u3 u3]". You then end up with a
lot of terms, but each of them has a closed-form solution.
The key to the variable change is an eigensystem analysis of the matrix
{{ -a, -d12/2, -d13/2 },
{ -d12/2, -b, -d23/2 },
{ -d13/2, -d23/2, -c }}
and the integral converges whenever this matrix is positive-definite.
Scott
--
Scott Hemphill hemphill at alumni.caltech.edu
"This isn't flying. This is falling, with style." -- Buzz Lightyear