I use the following definition of principal value: \begin{equation} \mathcal{P}\left(\frac{1}{f}\right)\left[h\right]=\lim_{\epsilon\to0}\sum_{x_i|f(x_i)=0}\left[\int_{-\infty}^{x_i-\epsilon}dx\frac{h(x)}{f(x)}+\int_{x_i+\epsilon}^{+\infty}dx\frac{h(x)}{f(x)}\right] \end{equation} where $h$ is a test function. With this definition I would like to compute \begin{equation} \mathcal{P}\left(\frac{1}{\lvert\vec{x}-\vec{x}'\rvert^3}\right) \end{equation} I am not sure how to compute it by using that definition. The ultimate goal for me is to compute this integral: \begin{equation} \int d^3x'\sqrt{g}\,\mathcal{P}\left(\frac{1}{\lvert\vec{x}-\vec{x}'\rvert^3}\right)\left(\frac{1}{\lvert \vec{x}'\rvert^n}\right) \end{equation} where $n$ is a non negative integer and $g$ is the metric for a 3 dimensional flate space. I could imagine to use spherical coordinates. Then I would have the integral: \begin{equation} 2\pi\int dr'd\theta'\,r'^2\sin\theta'\mathcal{P}\left(\frac{1}{(r^2+r'^2-2rr'\cos\theta)^{3/2}}\right)\frac{1}{r^{'n}} \end{equation} but I don't know how to continue. Any help is appreciated.
Asked
Active
Viewed 28 times
1
-
It's unclear what precisely you're trying to integrate. The expression with $\int d^3x'$ has a $\mathcal{P}$ inside the integral? And/or doesn't match the notation you begin with? – Sal Dec 14 '21 at 09:45
-
Yes, it is inside. Inside the integral I have $\mathcal{P}\left(\frac{1}{\lvert \vec{x}-\vec{x}'\rvert^3}\right)$ times $\left(\frac{1}{\lvert \vec{x}'\rvert^n}\right)$. I am not sure how I compute $\mathcal{P}\left(\frac{1}{\lvert \vec{x}-\vec{x}'\rvert^3}\right)$ using the definition – Ruben Campos Delgado Dec 14 '21 at 09:49
-
Is $g$ just $1$ here ? – LL 3.14 Dec 14 '21 at 09:52
-
You might be interested by my answer here https://math.stackexchange.com/questions/3723136/the-fourier-transform-of-1-p3/3724502#3724502 – LL 3.14 Dec 14 '21 at 09:59
-
ok, I have a look at it. Anyway, yes, in cartesian coordinates g should be just 1. – Ruben Campos Delgado Dec 14 '21 at 10:09
-
Notice also that your integral is a convolution so you might want to use the Fourier transform to solve it (or in certain cases just to give it a meaning) ... In particular if $0<a,n<3$, $|x|^{-a} * |x|^{-n} = c,|x|^{d-(a+n)}$. But here this is a more critical case and the result might depend on the value of $n$ ... – LL 3.14 Dec 14 '21 at 10:09