Find the gradient of 1/mod{r-r'}

  • Thread starter Thread starter Ashiataka
  • Start date Start date
  • Tags Tags
    Gradient
Join the discussion
Ask a follow-up here, or get your own question answered by working scientists, mathematicians and engineers — people, not an autocomplete.
Real named experts · corrections over time · the nuance an AI answer skips
8 replies · 14K views
Ashiataka
Messages
21
Reaction score
1

Homework Statement


Find [tex]\nabla\left( \dfrac{1}{\left| \vec{r}-\vec{r'}\right| }\right)[/tex]

Homework Equations





The Attempt at a Solution

[tex] \left| \vec{r}-\vec{r'}\right| =\sqrt{(x-x^\prime)^2 + (y-y^\prime)^2 + (z-z^\prime)^2}[/tex]
and so therefore the derivative of the scalar would be 0. Of course then it seems trivial. So perhaps I need to use the chain rule, but I don't know what to make my substitution for. I can't seem to escape the fact that the bottom always ends up being a scaler.

So I seem a bit stuck. Any advice on where to go onwards would be appreciated. Thank you.
 
on Phys.org
I don't understand how you get 0. What do you get for
[tex]\frac{\partial}{\partial x}((x-x^\prime)^2 + (y-y^\prime)^2 + (z-z^\prime)^2)^{-1/2}[/tex]?
Yes, you need to use the chain rule. You can break it up as finely as you like, but a reasonable choice would be
[tex]\frac{\partial f}{\partial x} \frac{\partial f^{-1/2}}{\partial f}[/tex]
where [tex]f = (x-x^\prime)^2 + (y-y^\prime)^2 + (z-z^\prime)^2[/tex]
 
Thank you.

Okay, so doing that I get [tex] \frac{\partial}{\partial x}\left(\frac{1}{\left| \vec{r}-\vec{r'}\right|}\right) = \frac{x^{\prime}-x}{\sqrt{(x-x^\prime)^2 + C}^3}[/tex]

by treating x' as a constant.Does that seem along the right path?
 
Sort of. I assume C here stands for (y-y')2 + (z-z')2, but why are you choosing to represent that with C? Suggests to me you are thinking of it as a constant, and so playing a different role from the (x-x')2 part. Yes, you had treat it as a constant for the purposes of ∂/∂x, but having done that it tells you the value of the derivative at any point (x,y,z), so they should still be considered variables on an equal footing. So I would suggest representing the whole square root part using the original modulus formula involving r.
What do you then get for the whole grad (∇) vector?
 
I got this:

[tex]\nabla \left(\frac{1}{\left| \vec{r}-\vec{r'}\right|}\right) = \frac{ (x^\prime-x) \vec{i} + (y^\prime-y) \vec{j} + (z^\prime-z)\vec{k}} {\sqrt{(x-x^\prime)^2 + (y-y^\prime)^2 + (z-z^\prime)^2}^3}[/tex]

by adding the x, y and z parts of the same form as above.
 
Last edited:
Two things wrong with that. First, I assume you meant ∇ on the left, not ∂/∂x. Secondly, you shouldn't add the three terms up. ∇ is a vector, namely (∂/∂x, ∂/∂y, ∂/∂z), so the answer should be a vector consisting of those three terms.
 
Yes, just typos. Corrected as above.
 
Thank you for your assistance :).
 
  • Like
Likes   Reactions: Shair