2
$\begingroup$

In multivariable calculus the line integrals was parameterized and denoted: $$ \int_C \mathbf{F} \bullet \, d\mathbf{r}=\int_D\mathbf{F}(\mathbf{r}(t)) \bullet \frac{d \mathbf{r}(t)}{dt} \, dt $$ However in electromagnetism the line integrals are confusing. The electric field: $$ \mathbf{E}(\mathbf{r}(t))=\frac{1}{4\pi \epsilon_0} \int_{path} \frac{\lambda(\mathbf{r}')}{\lvert \mathbf{r}-\mathbf{r'} \rvert^2} \frac{\mathbf{r}-\mathbf{r'}}{\lvert \mathbf{r}-\mathbf{r'} \rvert} \, dl' $$ $\mathbf{r}$ = field point and $\mathbf{r'}$ = source point of the charge.

What is the difference between this two integrals? Is the second a line integral along a vector field?

In math the procedure was; parametrize the curve, take the derivative of it, dot it with the parameterized field.

Should a parametrize the electric field or not?

I also have problem setting up the differential line element $dl'$. What is $d\mathbf{r}$ equivalent to in the second integral?

$\endgroup$
1
$\begingroup$

$dl'$ is equivalent to "$d|\mathbf{r}|$", it is essentially a "scalar length measure".

The electrodynamics integral you wrote here is a vector-valued integral, so no dotting happens. If you use a linear coordinate system, it may be evaluated as three scalar line integrals, one for each coordinate. Vector valued integrals cannot really be evaluated using a curved coordinate system.

Your integral should be evaluated as follows: $\mathbf{r}'$ is actually a curve $\mathbf{r}'(t)$ (I am not going to use ' as a notation for derivatives), and your integral is $$\mathbf{E}(\mathbf{r})=\kappa\int_\gamma\frac{\lambda(\mathbf{r}'(t))}{|\mathbf{r}-\mathbf{r}'(t)|^2}\frac{\mathbf{r}-\mathbf{r}'(t)}{|\mathbf{r}-\mathbf{r}'(t)|}\left|\frac{d\mathbf{r}'(t)}{dt}\right|dt. $$

Note that whilst I do not know the full context, from the form of the integral I can infer that only $\mathbf{r}'$ is parametrized by $t$. You are asking the value of $\mathbf{E}$ at a single point ($\mathbf{r}$), and this value is given by integrating the $\mathbf{r}'$ variable over a curve.


Edit: To answer your comment, you are almost correct.

Now that I know the full case, you also need to parametrize the $\mathbf{r}$ variable, but only because you are interested in values of $E$ solely on the $z$ axis!

Let us calculate the relevant vector quantities:

$$\mathbf{r}=\mathbf{r}(z)=z\mathbf{e}_z$$ (I am using the generic $z$ as a parameter instead of $b$, since I think this way it is clearer.) $$\mathbf{r}'(t)=a\cos(t)\mathbf{e}_x+a\sin(t)\mathbf{e}_y, $$ $$ \mathbf{r}(z)-\mathbf{r}'(t)=-a\cos(t)\mathbf{e}_x-a\sin(t)\mathbf{e}_y+z\mathbf{e}_z, $$ $$|\mathbf{r}(z)-\mathbf{r}'(t)|^2=a^2+z^2, $$ $$ \frac{d\mathbf{r}'(t)}{dt}=-a\sin(t)\mathbf{e}_x+a\cos(t)\mathbf{e}_y, $$ $$\left|\frac{d\mathbf{r}'(t)}{dt}\right|=a. $$

Now the integral is (since $\lambda$ is constant) $$ \mathbf{E}(\mathbf{r}(z))=\frac{\lambda}{4\pi\epsilon_0}\int_\gamma\frac{1}{a^2+z^2}\left(-\frac{a\cos(t)}{\sqrt{a^2+z^2}}\mathbf{e}_x-\frac{a\sin(t)}{\sqrt{a^2+z^2}}\mathbf{e}_y+\frac{z}{\sqrt{a^2+z^2}}\mathbf{e}_z\right)a\ dt=\\\frac{\lambda}{4\pi\epsilon_0}\frac{a}{(a^2+z^2)^{3/2}}\int_0^{2\pi}z\ dt\mathbf{e}_z=\frac{\lambda}{2\epsilon_0}\frac{az}{(a^2+z^2)^{3/2}}\mathbf{e}_z, $$ where the $\mathbf{e}_x$ and $\mathbf{e}_y$ integrals are annihilated because $\sin(t)$ and $\cos(t)$ are integrated on their full period.

| cite | improve this answer | |
$\endgroup$
  • $\begingroup$ Great, thanks! Shouln't $\mathbf{r}$ be parameterized (a function of $t$), i.e. $\mathbf{r}=\mathbf{r}(t)$? $\endgroup$ – JDoeDoe Jul 16 '16 at 21:18
  • $\begingroup$ I had problem with this question: A uniform line charge density $\lambda$ forms a circle of radius $a$ in the $xy$-plane ($z = 0$), with the center at the origin. Find the electric field at point $(0,0,b)$. Solution: ($\mathbf{r'}(t))$ is the source point, not the derivative) I don't really know there to start. Is the curve the same as the source point $\mathbf{r'}(t)$? Should I parametrize the circle in the $xy$-plane? If $x^2+y^2=a^2$, let $x(t)=\cos t , y(t)=\sin t, t\to[0,2\pi]$. $\endgroup$ – JDoeDoe Jul 16 '16 at 21:19
  • $\begingroup$ The position vectors are (parameterized) $$ \mathbf{r}(t)=b{\mathbf{\hat{z}}} $$ $$ \mathbf{r'}(t)=a(\cos t \,{\mathbf{\hat{x}}} +\sin t \, {\mathbf{\hat{y}}}) $$ $$ \mathbf{r}(t)-\mathbf{r'}(t)= b{\mathbf{\hat{z}}} - a(\cos t \,{\mathbf{\hat{z}}}+\sin t \, {\mathbf{\hat{y}}}) $$ Is this right? I'm kind of lost here. $\endgroup$ – JDoeDoe Jul 16 '16 at 21:20
  • $\begingroup$ @JDoeDoe I edited my post. $\endgroup$ – Bence Racskó Jul 16 '16 at 22:00
  • $\begingroup$ Thanks, really appreciate it! :) For surface charge density, is this correct (in terms of parameters and excluding parameterization of $\mathbf{r}$) $$ \mathbf{E(r)}=\frac{1}{4\pi\epsilon_0} \iint_{S} \frac{\sigma(\mathbf{r}'(s,t))}{|\mathbf{r}-\mathbf{r}'(s,t)|^2} \frac{\mathbf{r}-\mathbf{r}'(s,t)}{|\mathbf{r}-\mathbf{r}'(s,t)|} \biggl \lvert\frac{\partial\mathbf{r}'(s,t)}{\partial s} \times \frac{\partial\mathbf{r}'(s,t)}{\partial t} \biggl \rvert \, ds \, dt $$ $\endgroup$ – JDoeDoe Jul 17 '16 at 6:41

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.