Interchanging diff. and integr. rigorously

  • Thread starter Thread starter LouisR
  • Start date Start date
LouisR
Messages
1
Reaction score
0
I have some problems with what a prof. asked us in class. This is a class of analysis that is dealing with measure theory and stuff like Fourrier transform:
I am really stuck with the following ( really have problem to explain the steps rigorously ):

(d/dy){g(x,y)} means differentiates(partially) g with respect to y
i is the usual square root of -1

1-) For this one, x is in R^n, y is in R, f(x) is a Schwarz function ( or if you prefer in the Schwarz class, so as consequence, f is integrable over R^n and it is infinitely differentiable). Could anyone gives me the rigorous steps for this please:

integral( (d/dy){e^(-i*x*y)*f(x) dx} ) = (d/dy){integral(e^(-i*x*y)*f(x) dx)}



2-)Here is a simpler case: Let [a,b] be a close interval in the real. Integration is from a to b. Suppose that f(x,y) and (d/dx){f(x,y)} are continuous on R^2. Then again what are the rigorous justifications for this:

(d/dy){integral(f(x,y)dx)} = integral((d/dy){f(x,y)dx})

A really big thanks in advance if anybody could explain me this.
 
Physics news on Phys.org


In order to interchange differentiation and integration rigorously, we must use the fundamental theorem of calculus and the Leibniz integral rule. Let's first consider the first problem:

1-) For this one, x is in R^n, y is in R, f(x) is a Schwarz function (or if you prefer in the Schwarz class, so as consequence, f is integrable over R^n and it is infinitely differentiable). Could anyone gives me the rigorous steps for this please:

To start, let's rewrite the integral using the Leibniz integral rule:
integral( (d/dy){e^(-i*x*y)*f(x) dx} ) = integral( (d/dy){e^(-i*x*y)} * f(x) dx) + integral( e^(-i*x*y) * (d/dy){f(x)} dx)

Now, we can use the fundamental theorem of calculus to evaluate the first integral on the right-hand side:
integral( (d/dy){e^(-i*x*y)} * f(x) dx) = e^(-i*x*y) * f(x) + C

Substituting this back into our original integral, we have:
integral( (d/dy){e^(-i*x*y)*f(x) dx} ) = e^(-i*x*y) * f(x) + C + integral( e^(-i*x*y) * (d/dy){f(x)} dx)

Using the Leibniz integral rule again, we can interchange the differentiation and integration:
integral( e^(-i*x*y) * (d/dy){f(x)} dx) = (d/dy){integral(e^(-i*x*y)*f(x) dx)}

Therefore, we have:
integral( (d/dy){e^(-i*x*y)*f(x) dx} ) = e^(-i*x*y) * f(x) + C + (d/dy){integral(e^(-i*x*y)*f(x) dx)}

Now, since we are dealing with a Schwarz function, we know that it is integrable over R^n and infinitely differentiable. This means that the constant C is equal to 0, and we can rewrite our integral as:
integral( (d/dy){e^(-i*x*y)*f(x) dx} ) = e^(-i*x*y) *
 

Similar threads

Back
Top