I came across an interesting problem and am wondering if there is a quick answer to it from someone who's done this before before I whip out the books of integral tables.

Analytic expressions from assorted works for the magnetic field from a circular current loop seem to be badly divergent and have strange artifacts. I am doing some simulations to design some field coils for an experiment (common) and I started off with a finite element sim which seems to work well:


and the max field matches up to analytical expressions given my # of windings, etc ($B_{max} = (4/5)^{3/2} \mu_0 N I / R $) to within 5-10%, great. The caveat, of course is when I increase the number of windings (will need over 1000 for the fields I need) and this takes forever in my notebook. I sought out analytic solutions from prior works and came across the following:

Both have similar expressions with slightly different conventions (only component of vector potential is $A_\varphi$ which can be expressed with elliptic integrals). In cylindrical coordinates ($\rho, \varphi, z$), where z is offset from the ring center, the form for $B_\rho$ in these works is:

$$ B_\rho = \frac{Cz}{2\alpha^2\beta} \frac{1}{\rho} \left[ (R^2 + \rho^2 + z^2) E(k^2) - \alpha^2 K(k^2) \right] $$

where R is the ring radius and,

$$ \alpha^2 = R^2 + \rho^2 + z^2 - 2R\rho \\ \beta^2 = R^2 + \rho^2 + z^2 + 2R\rho \\ k^2 = 1 - \alpha^2/\beta^2$$

It would make sense that the fields fall off with $\rho$, and this form is claimed to be the form for all space (indeed the NASA work claims it reduces to the correct forms in certain limits ($x, y << R$, $z=0$). However, even looking at that expression it's obvious that it will be divergent at $\rho = 0$ as the numerator does not go to zero with $\rho \rightarrow 0$. Indeed, when this is coded up and executed, the divergence near $\rho = 0$ is extreme and obvious:


Isolating for one ring, the divergence we would expect is clear, and there is this strange artifact that loops around (in red):


I am using the numerical scheme in the meantime, but I am curious as to what is going on here and will swing back around to this when I can. Any insights are appreciated, cheers.

EDIT: I am asking about the forms in the linked papers and why they are assumed to be correct in those works when they clearly violate $\nabla \cdot B = 0 $ near $\rho = 0$, as shown in the plots.

  • $\begingroup$ What is this question about? Should we check both articles or your numerical method? $\endgroup$ – Alex Trounev Apr 18 at 20:30
  • $\begingroup$ Hey, I'm asking about the forms in the articles. My numerical method (as stated) seems to be producing reasonable fields. I am simply plotting the forms in those articles and they clearly violate $\nabla \cdot B = 0 $ near $\rho = 0$ and result in strange artifacts. I can link my notebook but I am simply plotting the equations I linked and using np.ellipk1m, np.ellipe. $\endgroup$ – Steve Novakov Apr 18 at 21:27
  • $\begingroup$ Is it Mathematica notebook? $\endgroup$ – Alex Trounev Apr 18 at 21:39
  • $\begingroup$ Jupyter, I'll link it shortly, I should have done it originally but I can't at the moment. $\endgroup$ – Steve Novakov Apr 18 at 21:44
  • $\begingroup$ For what it's worth, enforcing $\nabla\cdot\mathbf{B}=0$ can be tricky in numerical simulations. Is that what's happening here, IDK; but it's often something that is overlooked when writing code. $\endgroup$ – Kyle Kanos Apr 18 at 21:47

I derived expressions $B_{\rho},B_z$ using expression $A_{\phi }$ in cylindrical coordinates. These expressions coincided with those given in the article Simple Analytic Expressions for the Magnetic Field of a Circular Current Loop- see eqs (24),(25). For the output, I used Mathematica 12. Figure 1 shows the magnetic field lines of the current loop at different scales fig1

  • $\begingroup$ Thanks for that, this points to problems in my packages. I will investigate shortly. $\endgroup$ – Steve Novakov Apr 19 at 23:17

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.