2
$\begingroup$

I'm trying to understand an integration over the surface of a sphere that is used in one of the articles I'm reading. I don't know why I can't understand it as it seems to be a pretty straightforward integration and I have been used to more complex math but anyway.

Let $\textbf{r} = \textbf{x - x'}$, $r = |\textbf{r}|$ and $\textbf{n} = \frac{\textbf{r}}{r}$ where $|.|$ is the euclidian norm. The goal is to calculate the following integral for $i, j \in \{1,2,3\}$ :

$$I = \int_{A(r)} n_{i}n_{j}\text{d}A$$

Where $A(r)$ denotes a spherical surface of radius $r$.

The way I've gone about it is to say that for a spherical surface of radius $r$ we have

$$\text{d}A = r\sin(\phi)\text{d}\phi\text{d}\theta$$

with $(\phi, \theta) \in [0,\pi]\times[0,2\pi]$ and since $r_{i}$ or $r_{j}$ are not dependent of the angles we should have

$$I = 4\pi r_{i}r_{j}$$

However the article I'm reading has the result

$$I = \frac{4\pi r^{2}}{3} \delta_{ij}$$

I've been thinking about it but can't seem to find my mistake.

Thanks for your help kind stranger :)

$\endgroup$
1
$\begingroup$

Given your choice of parameters, you seem to be parametrizing $A(r)$ via \begin{equation} (\phi,\theta) \mapsto (r\cos\theta\sin\phi, r\sin\theta\sin\phi, r\cos\phi). \end{equation} This allows us to parametrize the unit normal vector $\mathbf{n}$: \begin{equation} \mathbf{n}(\phi,\theta) = (\cos\theta\sin\phi,\sin\theta\sin\phi,\cos\phi). \end{equation} From this we have \begin{align*} n_1n_{1} &= \cos^2\theta\sin^2\phi\\ n_1n_2 &= \sin\theta\cos\theta\sin^2\phi\\ n_1n_3 &= \cos\theta\sin\phi\cos\phi\\ n_2n_2 &= \sin^2\theta\sin^2\phi\\ n_2n_3 &= \sin\theta\sin\phi\cos\phi\\ n_3n_3 &= \cos^2\phi. \end{align*} We then have \begin{equation} I_{ij} = \int_0^{2\pi}\int_0^\pi n_in_j r^2\sin\phi d\phi d\theta. \end{equation} (Notice that the integrating factor should be $dA=r^2\sin\phi d\phi d\theta$. You can get this by computing the relevant Jacobian determinant for your parametrization.) We can then directly compute the six integrals. For instance, \begin{align*} I_{11} &= \int_0^{2\pi}\int_0^\pi r^2\cos^2\theta\sin^3\phi d\phi d\theta = r^2\left(\int_0^{2\pi}\cos^2\theta d\theta\right)\left(\int_0^\pi \sin^3\phi d\phi\right)\\ &= r^2\cdot \pi\cdot \frac{4}{3} = \frac{4\pi r^2}{3}\delta_{11}. \end{align*} The other integrals will work out similarly. The real problem seems to have been the integrating factor. Notice that if $i\neq j$, then $n_in_j$ has an odd-power $\theta$ term, and this will kill the integral.

$\endgroup$
  • $\begingroup$ Thanks a lot for your answer ! Any chance you would have a reference (undergrad or grad course) for this type of surface parametrization ? It looks like the formal math education I received overlooked curves, surfaces and volumes parametrizations $\endgroup$ – Experience111 Jun 20 '17 at 15:52
  • 1
    $\begingroup$ @JeremyDiallo Surface integrals (and the parametrizations necessary for them) are usually covered in an undergraduate multivariable calculus class. Most standard calculus textbooks should have what you need. $\endgroup$ – 211792 Jun 20 '17 at 16:14

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.