1. Limited time only! Sign up for a free 30min personal tutor trial with Chegg Tutors
    Dismiss Notice
Dismiss Notice
Join Physics Forums Today!
The friendliest, high quality science and math community on the planet! Everyone who loves science is here!

Homework Help: Magnetostatics: Magnetic Vector Potential

  1. May 3, 2010 #1
    1. The problem statement, all variables and given/known data
    Give an expression for the magnetic field and show that a magnetic vector exists such as [tex]\vec{A}(P) = A(r)\hat{z}[/tex] and [tex]\vec{B}(P) = \vec{\nabla} \times \vec{A}[/tex]
    For the infinite wire shown in figure 1.

    Here is a link to the figure and problem statement. The problem is the second problem on the first page.


    2. Relevant equations

    [tex] \vec{\nabla} \times \vec{A} = \left(\frac{1}{r}\frac{\partial A_{z}}{\partial \theta} - \frac{\partial A_{\theta}}{\partial z}\right)\hat{r} + \left(\frac{\partial A_{r}}{\partial z} - \frac{\partial A_{z}}{\partial r}\right)\hat{\theta} + \frac{1}{r}\left(\frac{\partial(rA_{\theta})}{\partial r} - \frac{\partial A_{r}}{\partial \theta}\right)\hat{z}[/tex] curl in polar coordinates

    [tex]\oint \vec{B} \cdot d\vec{l} = \mu_{0}I_{enc}[/tex] Ampere's Law

    3. The attempt at a solution

    First we must solve for the magnetic field at point P. Using Ampere's Law we get,

    [tex] \vec{B}(P) = \frac{\mu_{0}I}{2\pi r}\hat{\theta} [/tex] where the positive theta direction is into the page.

    Now in order to find A we can set the curl equations equal to the magnetic field equation by curl(A) = B.

    Doing so gives,

    [tex]\frac{1}{r}\frac{\partial A_{z}}{\partial \theta} - \frac{\partial A_{\theta}}{\partial z} = 0 [/tex]

    [tex]\frac{\partial A_{r}}{\partial z} - \frac{\partial A_{z}}{\partial r} = \frac{\mu_{0}I}{2\pi r}[/tex]

    [tex]\frac{\partial(rA_{\theta})}{\partial r} - \frac{\partial A_{r}}{\partial \theta} = 0[/tex]

    The answer to this problem (given to us by our instructor) is,

    [tex]\vec{A} = \left(-\frac{\mu_{0}I}{2\pi}ln(r) + K \right)\hat{z}[/tex]

    and the only way I can get there is by saying that because A points most often in the direction of the current (which is the z direction in this case) then,

    [tex]-\frac{dA_{z}}{dr} = \frac{\mu_{0}I}{2\pi r}.[/tex]

    Then solving this differential equation gives the desired result.

    Can someone please point me in the right direction as to how solve this problem in a more rigorous manner?

    Do I need to use the equation [tex]\vec{A} = \frac{\mu_{0}}{4\pi}\int \frac{\vec{J}(\vec{r'})}{r}d\tau '[/tex]?

    Many Thanks in advance,

    Last edited by a moderator: Apr 25, 2017
  2. jcsd
  3. May 3, 2010 #2
    You already know that:

    [tex]A_r = 0[/tex]


    [tex]A_{\theta} = 0[/tex]


    [tex]A_z = A(r)[/tex]

    since it was given to you in the problem. This simplifies your 3 diff eq's quite a bit.
  4. May 4, 2010 #3
    Thanks for your reply nickjer. So because A is a function of only r then all other derivatives go away and because it is only in the z direction then all other components are equal to zero. This then allows me to let [tex]-\frac{dA_{z}}{dr} = \frac{\mu_{0}I}{2\pi r}.[/tex] Is there anything else I must say for that condition to be satisfied?

    Thanks again,
  5. May 4, 2010 #4


    User Avatar
    Gold Member

    nope that does it:

    [tex]\vec{B}(P) = \vec{\nabla} \times \vec{A}[/tex]

    [tex]\vec{B}(P) = \frac{\mu_{0}I}{2\pi r}\hat{\theta} [/tex]

    [tex]\vec{\nabla} \times \vec{A}=-\frac{dA_{z}}{dr} [/tex]

    thus [tex]-\frac{dA_{z}}{dr}=\frac{\mu_{0}I}{2\pi r}\hat{\theta}[/tex]
  6. May 4, 2010 #5
    Thanks! I really appreciate the help kreil and nickjer.

  7. May 4, 2010 #6
    I now have a question concerning the third problem on that page attached.

    Problem Statement:

    We consider now two wires of axis (O,z) and separated by the distance (2a). The currents in the two wires are [tex]+I[/tex] and [tex]-I[/tex] (see figure). Show that the expression of the magnetic potential vector [tex]\vec{A}(P)[/tex] is given by

    [tex]\vec{A}(P) = \frac{\mu_{0}I}{2\pi}ln\left(\frac{r_{1}}{r_{2}}\right)\hat{z}[/tex]

    (the observation point is located at [tex]P(r, \theta, z=0)[/tex]). Note: The constant of integration is integration obtained by taking [tex]\vec{A}(O) = 0[/tex].

    Relevant Equations:

    Same as previous problem.

    Attempt at a solution:

    For this problem can I say the same as before? He did say that our answer should be of the form [tex]\vec{A}(P) = A(r)\hat{z}.[/tex]

    Thanks in advance.
    Last edited: May 4, 2010
  8. May 4, 2010 #7
    You solved for the vector potential from a single wire. So for two wires you will just add the vector potentials from each. But be careful with the direction of each of the vectors.
  9. May 4, 2010 #8
    I know from just looking at the answer that the -I current must be in the positive z-direction and the +I current must be in the negative z-direction but I don't know why.
  10. May 4, 2010 #9
    In the first problem you solved [itex]\vec{A}[/itex] for a current going in the +z direction. So use that to solve this next problem.
  11. May 4, 2010 #10
    Thanks for your reply nickjer.

    I think I got it. So because +I flows in the positive direction it must be negative and the opposite must be true for the wire with -I.

    One more question:

    For problem 4 on that document it says,

    Problem Statement
    The wires are now very close to each other. Determine [tex]\vec{A}(r,\theta)[/tex] such as r >> a and by keeping the first and second order terms in (a/r).

    Relevant Equations
    Remember that [tex](1 + \epsilon)^{n} = 1 + n\epsilon[/tex]

    and [tex]ln(1 + \epsilon) = \epsilon, \epsilon << 1[/tex]

    Attempt at a solution

    Now I know he wants us to approximate the answer by expanding[tex]\vec{A}(P) = \frac{\mu_{0}I}{2\pi}ln\left(\frac{r_{1}}{r_{2}}\right)[/tex]

    with a Taylor Expansion in [tex] \epsilon = \frac{a}{r} << 1[/tex] but I don't see how to do this when the distance between the wires [tex]2a[/tex] never comes up in any of our answers.

    I really appreciate all the help nickjer.
  12. May 4, 2010 #11
    Now you will need to write out [itex]r_1[/itex] and [itex]r_2[/itex] in terms of r and a. I suggest using the law of cosines.
  13. May 8, 2010 #12
    OK. Thanks again for all of your help nickjer.
Share this great discussion with others via Reddit, Google+, Twitter, or Facebook