Gaussian integral trouble

In summary, the conversation discusses a mathematical problem involving integrals and their derivatives. The goal is to derive a formula for the integral of x^4 multiplied by an exponential function. Two different approaches are discussed, but one leads to a contradiction. The conversation suggests using a substitution to simplify the problem.
  • #1
AdrianMay
121
4
Hi folks,

I'm trying to get from the established relation:

$$ \int_{-\infty}^{\infty} dx.x^2.e^{-\frac{1}{2}ax^2} = a^{-2}\int_{-\infty}^{\infty} dx.e^{-\frac{1}{2}ax^2} $$

to the similarly derived:

$$ \int_{-\infty}^{\infty} dx.x^4.e^{-\frac{1}{2}ax^2} = 3a^{-4} \int_{-\infty}^{\infty} dx.e^{-\frac{1}{2}ax^2} $$

but instead I'm winding up with:

$$ \int_{-\infty}^{\infty} dx.x^4.e^{-\frac{1}{2}ax^2} = (4a^{-3} - a^{-4}) \int_{-\infty}^{\infty} dx.e^{-\frac{1}{2}ax^2} $$.

Evidently the difference is that I have ##a^{-3}## where I should have ##a^{-4}## but I can't seem to fault my own logic. First I differentiate the thing I started with:

$$ -2\frac{\partial}{\partial a} [ \int_{-\infty}^{\infty} dx.x^2.e^{-\frac{1}{2}ax^2} ] = -2\frac{\partial}{\partial a} [ a^{-2}\int_{-\infty}^{\infty} dx.e^{-\frac{1}{2}ax^2} ] $$

apply the chain rule:

$$ \int_{-\infty}^{\infty} dx.x^4.e^{-\frac{1}{2}ax^2} = -2 \{ \frac{\partial a^{-2}}{\partial a}.\int_{-\infty}^{\infty} dx.e^{-\frac{1}{2}ax^2} + a^{-2}.\frac{\partial}{\partial a}\int_{-\infty}^{\infty} dx.e^{-\frac{1}{2}ax^2} \} $$

and hit the problem in what looks like the easy bit:

$$ \int_{-\infty}^{\infty} dx.x^4.e^{-\frac{1}{2}ax^2} = 4a^{-3}.\int_{-\infty}^{\infty} dx.e^{-\frac{1}{2}ax^2} - a^{-4}.\int_{-\infty}^{\infty} dx.e^{-\frac{1}{2}ax^2} $$

(where the last term follows from the relation I started with.)

So where's the bug?

Thanks in advance,
Adrian.
 
Last edited:
Physics news on Phys.org
  • #2
I can get the correct result as well. Start with:

$$ \int_{-\infty}^{\infty} dx.e^{-\frac{1}{2}ax^2} = \sqrt{2\pi}a^{-\frac{1}{2}} $$

double diff right away:

$$ 4\frac{\partial^2}{\partial a^2} \int_{-\infty}^{\infty} dx.e^{-\frac{1}{2}ax^2} = 4\frac{\partial^2}{\partial a^2} \sqrt{2\pi}a^{-\frac{1}{2}} $$

$$ \int_{-\infty}^{\infty} dx. 4\frac{\partial^2}{\partial a^2} e^{-\frac{1}{2}ax^2} = 4\sqrt{2\pi}.\frac{-1}{2}.\frac{-3}{2}.a^{-\frac{5}{2}} $$

$$ \int_{-\infty}^{\infty} dx. x^4. e^{-\frac{1}{2}ax^2} = 3a^{-2}\sqrt{2\pi}.a^{-\frac{1}{2}} = 3a^{-2}\int_{-\infty}^{\infty} dx.e^{-\frac{1}{2}ax^2} $$

But now I'm looking at a contradiction.What was wrong with the first approach?
 
Last edited:
  • #3
OK, I figured it out. The starting point was supposed to be:

$$ \int_{-\infty}^{\infty} dx.x^2.e^{-\frac{1}{2}ax^2} = a^{-1}\int_{-\infty}^{\infty} dx.e^{-\frac{1}{2}ax^2} $$

so I'd get to:

$$ \int_{-\infty}^{\infty} dx.x^4.e^{-\frac{1}{2}ax^2} = -2 \{ \frac{\partial a^{-1}}{\partial a}.\int_{-\infty}^{\infty} dx.e^{-\frac{1}{2}ax^2} + a^{-1}.\frac{\partial}{\partial a}\int_{-\infty}^{\infty} dx.e^{-\frac{1}{2}ax^2} \} $$

$$ \int_{-\infty}^{\infty} dx.x^4.e^{-\frac{1}{2}ax^2} = -2 \{ -a^{-2}.\int_{-\infty}^{\infty} dx.e^{-\frac{1}{2}ax^2} + a^{-1}.\int_{-\infty}^{\infty} dx.\frac{\partial}{\partial a}e^{-\frac{1}{2}ax^2} \} $$

$$ \int_{-\infty}^{\infty} dx.x^4.e^{-\frac{1}{2}ax^2} = -2 \{ -a^{-2}.\int_{-\infty}^{\infty} dx.e^{-\frac{1}{2}ax^2} + a^{-1}.-\frac{1}{2}.\int_{-\infty}^{\infty} dx.x^2.e^{-\frac{1}{2}ax^2} \} $$

$$ \int_{-\infty}^{\infty} dx.x^4.e^{-\frac{1}{2}ax^2} = -2 \{ -a^{-2}.\int_{-\infty}^{\infty} dx.e^{-\frac{1}{2}ax^2} + a^{-1}.-\frac{1}{2}.a^{-1}\int_{-\infty}^{\infty} dx.e^{-\frac{1}{2}ax^2} \} $$

$$ = 3a^{-2} \int_{-\infty}^{\infty} dx.e^{-\frac{1}{2}ax^2} $$

as expected. Now for that funky matrix stuff, which will doubtless lead me back here.
 
  • #4
One way to avoid some of the trouble would be to substitute
y2 = ax2

Then the power of a could be taken outside the integral.
 

What is a Gaussian integral?

A Gaussian integral is a type of integral that involves the Gaussian function, which is a mathematical function that describes the normal distribution of a set of data. It is commonly used in statistics, probability, and physics.

Why is it called a "Gaussian" integral?

The name comes from the German mathematician and physicist, Carl Friedrich Gauss, who first described the function in the early 19th century. He used it to analyze astronomical data and it has since been used in various fields of science.

What makes Gaussian integrals difficult to solve?

Gaussian integrals can be difficult to solve because they involve complex numbers and can have infinite limits. They also often require advanced mathematical techniques such as contour integration or series expansions.

What are some real-world applications of Gaussian integrals?

Gaussian integrals have many practical applications, such as in probability and statistics to model the distribution of data. They are also used in physics to solve problems in electromagnetism, quantum mechanics, and thermodynamics.

Are there any shortcuts or tricks for solving Gaussian integrals?

Yes, there are various techniques that can be used to simplify and solve Gaussian integrals, such as using symmetry properties, substitution, and the method of residues. However, the best approach may depend on the specific integral being solved.

Similar threads

Replies
3
Views
695
Replies
2
Views
1K
  • Calculus
Replies
4
Views
1K
Replies
19
Views
3K
Replies
3
Views
1K
Replies
3
Views
1K
Replies
1
Views
927
Replies
2
Views
938
Replies
5
Views
1K
  • Calculus
Replies
9
Views
1K
Back
Top