Suppose that $\Omega$ is a bounded domain in $\mathbb{R}^3$ with $C^1$ boundary. I want to prove that $$\int_{\partial \Omega} \nu(y) \cdot \frac{y}{|y|^3} \, dS(y)=\left\{\begin{array}{cc}0 & 0 \in \mathbb{R}^3 \setminus \overline{\Omega} \\ 4\pi & 0 \in \Omega \end{array}\right.$$ where $\nu(y)$ is the unit exterior normal, and $dS$ is surface measure on $\partial\Omega$. If $0 \in \mathbb{R}^3\setminus\overline{\Omega}$ we'll use the divergence theorem applied to the $C^1$ vector field $F(y)=y/|y|^3$. In the case $0 \in \Omega$, we'll do the same thing to the vector field $y/|y|^3 \in C^1(\Omega\setminus\overline{B(0,\varepsilon)})$ for $\varepsilon>0$ sufficiently small. In either case, the result should follow from $\operatorname{div} F=0$, but I am unable to show this. Can someone help?


First notice that $y/|y|^3=\nabla (|y|^{-1})$ so that your integrand is $\partial_\nu f$, with $f(y)=1/|y|$.

If $0\in \mathbb{R}^3\setminus \bar{\Omega}$, then $f$ is harmonic in $\Omega$ so that the divergence theorem gives $$ \int_{\partial \Omega} \partial_\nu f(y) dS = \int_\Omega \Delta f =0. $$ If $0\in \Omega$ and $\Omega_\varepsilon=\Omega\setminus B(0,\varepsilon)$, then $f$ is harmonic in $\Omega_\varepsilon$ so that by the divergence theorem $$ 0= \int_\Omega \Delta f = \int_{\partial \Omega_\varepsilon} \partial_\nu f(y) dS $$ The last integral is equal to $$ \int_{\partial \Omega} \partial_\nu f(y) dS - \int_{\partial B(0,\varepsilon)} \partial_\nu f(y) dy. $$ Now the integral over the boundary of the ball we can compute since $\nu(y)=y/\varepsilon$ and we get $$ \int_{\partial B(0,\varepsilon)} \partial_\nu f(y) dy= \int_{\partial B(0,\varepsilon)}\frac{1}{\varepsilon^2}= \frac{4\pi \varepsilon^2}{\varepsilon^2}=4\pi. $$ Combining all this, we get the result.

  • $\begingroup$ I was able to come up with roughly this solution assuming that $\operatorname{div} F=0$. But why is $\operatorname{div} F=0$? Sorry if this is trivial; analysis is not my area. $\endgroup$ Jan 13 '15 at 2:25
  • $\begingroup$ @dmdmdmdmdmd: Could you be more specific? Do you know the definition of $\text{div}F$? Do you know how to take partial derivatives? $\endgroup$
    – Jose27
    Jan 13 '15 at 4:42
  • $\begingroup$ I do. And you can disregard; I figured out I was making a silly mistake. $\endgroup$ Jan 13 '15 at 15:45

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.