I am trying to write a PIC (Particle In Cell) code to simulate plasma physics. I am starting with the simplest case, which is a 1D system with a longitudinal field $E = E(x)$. I am using this article as reference (it has some typos, though)

In it, finite differences are applied to Gauss Law, which gives:

$$E(x + dx, t) =dx \rho(x,t)/\epsilon_0 + E(x,t)$$

(The article uses another normalisation) I would expect the electric field to vanish as x increases. However, this equation seems to say that if $\rho > 0$, then $E(x + dx) > E(x), \forall x$. This would be the result even for the simplest case, i.e., a single point-charge. Where is the mistake? Thank you!

  • $\begingroup$ This question is equal to another, or am I wrong? Anyway what you say is valid in 3D, not in 1D $\endgroup$ – Mark_Bell Apr 11 at 23:57
  • $\begingroup$ @Mark_Bell It’s a delete-and-repost. $\endgroup$ – G. Smith Apr 12 at 0:00
  • $\begingroup$ @G.Smith yes, since the other one wasn’t clear enough. Should I put in the question that it is a delete-and-repost? Thank you $\endgroup$ – Ali Esquembre Kucukalic Apr 12 at 4:58
  • $\begingroup$ @Mark_Bell I think it is valid for longitudinal fields. In fact, take a look to equation (15) in the quoted link. $\endgroup$ – Ali Esquembre Kucukalic Apr 12 at 5:00
  • $\begingroup$ You can always edit a question to make it clearer. This is preferable to deleting and reposting. $\endgroup$ – G. Smith Apr 12 at 5:16

The charge density in a PIC code has a finite extension, i.e., $\rho(x) \neq 0$ if $x$ belongs to some interval $[x_p - \Delta x, x_p + \Delta x]$. Thus, the solution comes by imposing the initial conditions far away from the charge equal to zero, i.e.,

$E(x,t) = 0$ for $x << x_p$,

and then keep on updating $E(x)$ until we reach $x_p - \Delta x$ so that $\rho(x_p - \Delta x)\neq 0$. After we cover the interval in which the charge is not zero, the immediate next point will give $E(x_p + \Delta x + 2dx) = \rho(x_p + \Delta x + dx) + E(x_p + \Delta x) = E(x_p + \Delta x)$, i.e., a uniform field which is the correct behavior in a 1D system, where $E(x)$ goes with $1/r^0$.


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.