0

Recently, I've made a question regarding the proof of $\nabla ^2 G(\textbf{r},\textbf{r}') = \delta (\textbf{r}-\textbf{r}')$ for $G(\textbf{r},\textbf{r}')=\frac{1}{4\pi|\textbf{r}-\textbf{r}'|}$. This question was answered and can be found here https://math.stackexchange.com/a/3816558/799584.

However, I'm now wondering whether or not we could affirm that

$$\frac{d^2}{dx^2} G(\textbf{r},\textbf{r}') = \frac{1}{3} \delta (\textbf{r}-\textbf{r}')$$

I tend to believe that this should be right as all $x,y$ and $z$ coordinates should be equivalent. But is there any formal proof for that?

Thanks!

J.-E. Pin
  • 40,163
sined
  • 377

1 Answers1

1

Since $ \partial_x^2 G = \nabla\cdot(\partial_xG,0,0) = \nabla\cdot(\partial_xG \,\hat{x}) $ we get $$ \iiint_{\Omega} \partial_x^2 G \, dV = \iiint_{\Omega} \nabla\cdot(\partial_xG \,\hat{x}) \, dV = \iint_{\partial\Omega} (\partial_xG \,\hat{x}) \cdot n\,dS = \hat{x}\cdot\iint_{\partial\Omega} \partial_xG \, n\,dS . $$

Taking $\Omega=B_r(\mathbf{0})$ makes $$ \iint_{\partial\Omega} \partial_xG \, n\,dS = - \iint_{S_r(\mathbf{0})} \frac{x}{|\mathbf{r}|^3} \frac{\mathbf{r}}{|\mathbf{r}|} \, dS = - \iint_{S_r(\mathbf{0})} \frac{(x^2,xy,xz)}{|\mathbf{r}|^4} \, dS . $$ Here, $\iint_{S_r(\mathbf{0})} \frac{xy}{|\mathbf{r}|^4} \, dS$ and $\iint_{S_r(\mathbf{0})} \frac{xz}{|\mathbf{r}|^4} \, dS$ will vanish because of the integrands are odd and the domain of integration is symmetric. So only $\iint_{S_r(\mathbf{0})} \frac{x^2}{|\mathbf{r}|^4} \, dS$ will contribute. Changing to spherical coordinates we get $$ \iint_{S_r(\mathbf{0})} \frac{x^2}{|\mathbf{r}|^4} \, dS = \int_{0}^{2\pi} d\phi \int_{0}^{\pi} d\theta \, r^2 \sin\theta \frac{r^2 \sin^2\theta \cos^2\phi}{r^4} = \int_{0}^{2\pi} d\phi \cos^2\phi \int_{0}^{\pi} d\theta \, \sin^3\theta = \pi \cdot \frac{4}{3} = \frac{4\pi}{3} . $$ Thus we have $ \iiint_{B_r(\mathbf{0})} \partial_x^2 G \, dV = -\frac{4\pi}{3} $ independent of $r$ so we have a term $-\frac{4\pi}{3}\delta(\mathbf{r})$ in $\partial_x^2 G.$

But we should also remember that outside of origin, we have $$ \partial_x^2 G = \partial_x \left(\partial_x \frac{1}{|\mathbf{r}|} \right) = \partial_x \left(- \frac{x}{|\mathbf{r}|^3} \right) = -\frac{1\cdot|\mathbf{r}|^3 - x\cdot 3|\mathbf{r}|^2\cdot x/|\mathbf{r}|}{|\mathbf{r}|^6} \\ = -\frac{|\mathbf{r}|^4 - 3|\mathbf{r}|^2x^2}{|\mathbf{r}|^7} = -\frac{|\mathbf{r}|^2 - 3x^2}{|\mathbf{r}|^5} = -\frac{-2x^2+y^2+z^2}{|\mathbf{r}|^5} . $$

Without having made certain that nothing has been missed, I therefore claim that $$ \partial_x^2 G(\mathbf{r}) = \frac{2x^2-y^2-z^2}{|\mathbf{r}|^5} - \frac{4\pi}{3}\delta(\mathbf{r}) . $$

md2perpe
  • 26,770
  • Would you mind commenting on this being a discontinuous function as different limits yield different values? – sined Sep 12 '20 at 19:58
  • @denis. What expression yields different values depending on how you take the limit? – md2perpe Sep 12 '20 at 20:02