1

how can I verify Ampere's law using Biot- Savart equation on a circular current:

enter image description here

namely, to prove:

$$\oint\limits_{W}B\,\mathrm{d}w=\mu_0 I$$

what is the curve W we should integrate with?

techie11
  • 429
  • Ampere's law applies to all closed curves, not just one in particular. So proving it by checking one specific curve $W$ is a nonstarter. (In principle one can switch to the differential form of Ampere's law $\nabla\times \vec{B}=\mu_0 \vec{J}$ and show that the curl of $\vec{B}$ vanishes except at the wire. But that requires knowing the B-field everywhere, not just along the axis, and thus is not terribly practical.) – Semiclassical Aug 04 '24 at 21:19
  • @Semiclassical I thought about the differential form but normally for a filiform wire there's no such as a current density, there's just $I$ so I think it has to be integral form. Maybe you could argue the density might be expressable using Dirac delta functions. – Hug de Roda Aug 04 '24 at 21:24
  • @Conreu You can indeed express $\vec{J}$ in that scenario in terms of Dirac delta functions, the same as you could for either the divergence of a point charge's E-field or the curl of an infinite line current. (See https://physics.stackexchange.com/questions/710553/what-is-the-current-density-of-a-circular-current for specifics in this particular case.) But admittedly this is not a calculation I'd be willing to do myself. – Semiclassical Aug 04 '24 at 21:31
  • @Semiclassical understand that Ampere's law applies to any closed curves. But is there one such curve that can be used to verify the law easily? – techie11 Aug 04 '24 at 21:47
  • You can pick a curve to test it, sure. But that will not be a proof, any more than by picking some polynomial and fully factoring it is a proof of the fundamental theorem of algebra. – Semiclassical Aug 04 '24 at 22:05

1 Answers1

3

Assuming the filament is in magnetostatic conditions, Ampère's law reads $$\oint_{\mathrm W}\,\mathbf B\cdot\mathrm d\boldsymbol w=\mu_0 I_{\text{enc}}$$ where $\mathrm d\boldsymbol w$ is tangent to the curve $\mathrm W$.

Let's then prove how Biot-Savart's law satisfies Ampère's law for any curve:

Substituting Biot-Savart's law we get that $$\oint_{\mathrm W}\oint_C\dfrac{\mathrm d\boldsymbol\ell\times(\boldsymbol{r-r'})}{|\boldsymbol{r-r'}|^3}\cdot\mathrm d\boldsymbol w=4\pi$$ has to hold. Let's compute that integral. First we'll employ $$\boldsymbol\nabla\times\left(\dfrac{\mathbf a}{\phi}\right)=\boldsymbol\nabla\dfrac{1}{\phi}\times\mathbf a+\dfrac{1}{\phi}\boldsymbol\nabla\times\mathbf a$$ where we identify $\mathbf a=\mathrm d\boldsymbol \ell$ and $\phi=|\boldsymbol{r-r'}|$. Hence, since $\mathrm d\boldsymbol\ell$ only depends on $\boldsymbol r'$, $$\oint_{\mathrm W}\oint_C\dfrac{\mathrm d\boldsymbol\ell\times(\boldsymbol{r-r'})}{|\boldsymbol{r-r'}|^3}\cdot\mathrm d\boldsymbol w=\oint_{\mathrm W}\boldsymbol\nabla\times\oint_C\dfrac{\mathrm d\boldsymbol\ell}{|\boldsymbol{r-r'}|}\cdot\mathrm d\boldsymbol w$$ Using Stokes' theorem, the integral becomes $$\iint_{S_\mathrm W}\boldsymbol\nabla\times\boldsymbol\nabla\times\oint_C\dfrac{\mathrm d\boldsymbol\ell}{|\boldsymbol{r-r'}|}\cdot\mathrm d\boldsymbol S_\mathrm W$$ where $$\boldsymbol\nabla\times\boldsymbol\nabla\times\oint_C\dfrac{\mathrm d\boldsymbol\ell}{|\boldsymbol{r-r'}|}=\boldsymbol\nabla\left(\oint_C\mathrm d\boldsymbol\ell\cdot\boldsymbol\nabla\dfrac{1}{|\boldsymbol{r-r'}|}\right)-\boldsymbol\nabla^2\oint_C\dfrac{\mathrm d\boldsymbol\ell}{|\boldsymbol{r-r'}|}$$ but $$\oint_C\mathrm d\boldsymbol\ell\cdot\boldsymbol\nabla\dfrac{1}{|\boldsymbol{r-r'}|}=-\oint_C\mathrm d\boldsymbol\ell\cdot\boldsymbol\nabla'\dfrac{1}{|\boldsymbol{r-r'}|}=-\oint_C \mathrm d\left(\dfrac{1}{|\boldsymbol{r-r'}|}\right)=0$$ and so the first term vanishes. Moreover, $$\boldsymbol\nabla^2\dfrac{1}{|\boldsymbol{r-r'}|}=4\pi\delta^{(3)}(\boldsymbol{r-r'}),$$ and thus, the original integral reduces to $$\begin{aligned} &\oint_{\mathrm W}\oint_C\dfrac{\mathrm d\boldsymbol\ell\times(\boldsymbol{r-r'})}{|\boldsymbol{r-r'}|^3}\cdot\mathrm d\boldsymbol w\\ &=-4\pi\iint_{S_{\mathrm W}}\oint_C \delta^{(3)}(\boldsymbol{r-r'})\,\mathrm d\boldsymbol\ell\,\mathrm d\boldsymbol S_{\mathrm W}\\ &=-4\pi\iint_{S_{\mathrm W}}[\delta(y-y')\delta(z-z')\boldsymbol i+\delta(x-x')\delta(z-z')\boldsymbol j+\delta(x-x')\delta(y-y')\boldsymbol k]\cdot\mathrm d\boldsymbol S_{\mathrm W}=4\pi\,\blacksquare \end{aligned}$$

Hug de Roda
  • 4,060
  • One point here: the double-loop integral is Gauss's linking integral, and as such should count the number of times the curve $W$ links with the curve $C$. That this comes out as linking number $1$ in this case is presumably implicit in how $S_W$ is defined in the case of a single-linking, but that's not an altogether trivial point to pass over. (It might be better to frame the second-to-last integral as counting the number of times that the contour $C$ "pierces" $S_W$.) – Semiclassical Aug 05 '24 at 04:14
  • How to understand this is what we need to prove: $$\oint_{\mathrm W}\oint_C\dfrac{\mathrm d\boldsymbol\ell\times(\boldsymbol{r-r'})}{|\boldsymbol{r-r'}|^3}\cdot\mathrm d\boldsymbol w=4\pi$$ – techie11 Aug 05 '24 at 20:19
  • Where can I find the relationship between Laplacian and Dirac's delta function: $$\boldsymbol\nabla^2\dfrac{1}{|\boldsymbol{r-r'}|}=4\pi\delta^{(3)}(\boldsymbol{r-r'})$$ – techie11 Aug 05 '24 at 20:22
  • 1
    Biot-Savart law is $$\mathbf B=\dfrac{\mu_0 I}{4\pi}\oint_C \dfrac{\mathrm d\boldsymbol\ell\times(\boldsymbol{r-r'})}{|\boldsymbol{r-r'}|}$$ if you substitute this in Ampère's law you'll be able to cancel out the $\mu_0$ and $I$ on both sides and isolate the $4\pi$ to the other side.

    As for the Dirac delta relationship you sould address to this post: https://math.stackexchange.com/questions/368155/where-does-the-relation-nabla21-r-4-pi-delta3-bf-r-between-laplacian

    – Hug de Roda Aug 05 '24 at 23:52
  • I proved it for the general case because, as they said in the comment section above, it is really difficult to compute analytically these integrals even for $\mathrm W$ curves as simple as circles or squares. – Hug de Roda Aug 05 '24 at 23:54
  • I'm still trying to figure out an alternative to this proof because I think I made some mistakes and it might contain some flaws... – Hug de Roda Aug 05 '24 at 23:57