2
$\begingroup$

An electrostatic potential associated with some delocalized charge $\int \rho(\mathbf{r}) d{\mathbf{r}}$ is given by:

$$v_H(\mathbf{r}) = \int \frac{\rho(\mathbf{r'})}{|\mathbf{r}-\mathbf{r'}|}d\mathbf{r'}$$

This potential is finite at $\mathbf{r}=0$. Since $\frac{1}{|\mathbf{r}-\mathbf{r'}|}$ is a solution to a singular Poisson's equation, we can show that:

$$\nabla^2v_H(\mathbf{r})=-4\pi\rho(\mathbf{r})$$ where $\rho$ is a smooth function being finite everywhere.

Would a function

$$v(\mathbf{r}) = \int \frac{\rho(\mathbf{r'})}{|\mathbf{r}-\mathbf{r'}|^n}d\mathbf{r'}$$

(where $n$ is some nonnegative integer)

be also finite at $\mathbf{r}=0$?

In this case, $\frac{1}{|\mathbf{r}-\mathbf{r'}|^n}$ does not represent a solution to Poisson's equation ($\mathbf{r} \in \mathbb{R}^3$) and the above analysis is not valid.

$\endgroup$
  • $\begingroup$ Since you said yourself, it does not represent a solution to Poisson's equation, why would be interested it these functions? $\endgroup$ – resgh Nov 30 '12 at 0:43
  • $\begingroup$ I am working in the field of quantum chemistry. I would like to use a model potential like the one I am asking about in electronic structure calculations. $\endgroup$ – molkee Nov 30 '12 at 0:49
  • $\begingroup$ I don't see why this would NOT be finite at $r=0$... $\endgroup$ – Dylan O. Sabulsky Nov 30 '12 at 2:52
  • $\begingroup$ Let me see if I can't make some plots for this. $\endgroup$ – Dylan O. Sabulsky Nov 30 '12 at 2:56
  • 1
    $\begingroup$ nope, the question was inspired by some problem in DFT $\endgroup$ – molkee Nov 30 '12 at 3:32
1
$\begingroup$

Maybe too simple, and certainly not a general solution, but anyway here goes.

The simplest case is a spherical charge distribution:

$$\rho= \left\{ \begin{array}{cc} \rho_0 & r \le a \\ 0 & r>a \end{array} \right. $$ Then

$$ V(0) = \rho_0 \int_0^a \frac{4 \pi r^2}{r^n} dr = 4 \pi \rho_0 \int_0^a r^{2-n} dr $$

Cases:

$$ 0<n<3 \qquad V(0) = \frac{4 \pi \rho_0}{3-n} a^{3-n} \text{ , finite}$$

$$ n=3 \qquad V(0) = 4 \pi \rho_0 \lim_{\epsilon\rightarrow 0} \ln \left(\frac{r}{\epsilon} \right) \rightarrow \infty $$

$$ n>3 \qquad V(0) = \frac{4 \pi \rho_0}{n-3} \lim_{\epsilon\rightarrow 0} \left(\frac{1}{\epsilon^{n-3}} - \frac{1}{a^{n-3}} \right) \rightarrow \infty $$

In all these cases, the denominator of the original integral, $|r-r'|^n$, grows without bound for large $r$, and the charge is localized around the origin, so

$$ \lim_{r \rightarrow \infty} V(r) =0 $$

So, $n=3$ divides the finite from the infinite cases. (Technically, I suppose this calc constitutes a counter-example for $n \ge 3$ by exhibiting an unbounded result for those cases).

$\endgroup$

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.