# Regularizing a divergent integral with a test function or convergence factor

1. Feb 19, 2009

### graphicsRat

I'm trying to understand a paper in which the authors use a number of test functions (are they the same as convergence factors) to make integrate unintegrable functions. Now here is my ignorant question: why is this acceptable? The product of the original function and the test function or convergence factor surely is an entirely new function. Why is the integral of this new function the same as the integral of the old function which I recall was unintegrable? In layperson terms, can we be sure that the integral of the new function is valid and or correct?

Thanks

2. Feb 19, 2009

### tiny-tim

Hi graphicsRat!

It depends on the context …

usually this is done to a subsidiary integral in the middle of a much larger integration …

there are theorems that tell you whether it's justified or not …

doesn't the paper give any references?

3. Feb 19, 2009

### graphicsRat

Thanks Tim.

No the paper didn't give any references. Which theorems are you referring to? I'd like to look them up. What do you mean by a "subsidiary integral in the middle of a much larger integration"?

I'm sorry to ask so many annoying questions.

4. Feb 20, 2009

### tiny-tim

dunno … i wasn't really concentrating when they went over that boring stuff
in quantum field theory, there are really long integrals, and you can rearrange the order of the variables so as to get some easy integrals in the middle

(btw, this rearranging involves combinatorial additions of integrals, and Feynman diagrams are what keep track of the different combinations)

but those easy integrals are of functions (technically called "distributions", such as the Dirac delta "function" … see http://en.wikipedia.org/wiki/Distribution_(mathematics)) which don't actually converge on their own, because they oscillate, but are ok in the larger integral because the oscillations are "damped down" by the rest of the long integral