I have been studying Euler-Lagrange in Variation Calculus. I am comfortable with the formulation when the function under the integral is of the form f = f(x, y). But I am unsure as to how this change for a function given in polar coordinates f = f(r, theta)


$\begin{equation*} I\ ( u) \ =\ \int ^{1}_{0}\int ^{2\pi }_{0} F\left( r,\ \theta ,\ u,\ \frac{\partial u}{\partial r} ,\ \frac{\partial u}{\partial \theta }\right) \ rd\theta dr \end{equation*}$

so how would you calculate the variation, $\delta I $

Will it be

$ \delta I (u) = \int ^1_0 \int ^{2\pi}_0 (\frac {\partial F} {\partial u} + \frac {\partial F} {\partial u ^`_r} {\delta u ^`_r} + \frac {\partial F} {\partial u ^`_\theta} {\delta u ^`_\theta} ) rd\theta dr$

I am confused about this, this seems incorrect, what should be done with the $r$ at the end? Also how do I apply Gauss divergence theorem at the end to seperate the boundary conditions from the rest of the integral.


You can just define $G=F \cdot r$, use the standard methods (treating $r$ and $\theta$ the same as cartesian coordinates), and transform back to $F$ at the end.

| cite | improve this answer | |
  • $\begingroup$ Thank you so much. Doing as you say I would get the Euler-Lagrange equation and the following part $ \begin{equation*} \int ^{1}_{0}\int ^{2\pi }_{0}\frac{\partial }{\partial r}\left(\frac{\partial G}{\partial u^{'}_{r}} \delta u\right) \ +\ \frac{\partial }{\partial \theta }\left(\frac{\partial G}{\partial u^{'}_{\theta }} \delta u\right) \ d\theta dr \end{equation*}$ Can you also help me with how I can apply the Gauss divergence theorem on this to get the natural boundary conditions? $\endgroup$ – user1887239 Dec 30 '18 at 12:02
  • $\begingroup$ Does doing it the same as for cartesian coordinates not work? $\endgroup$ – Eddy Dec 30 '18 at 12:56
  • $\begingroup$ With cartesian I'd use Gauss divergence theorem at this point to get the natural BCs which are of the form $\begin{equation*} I\ ( u) \ =\ \int ^{1}_{0}\int ^{1}_{0}\frac{\partial }{\partial x}\left(\frac{\partial F}{\partial u^{'}_{x}} \delta u\right) \ +\ \frac{\partial }{\partial y}\left(\frac{\partial G}{\partial u^{'}_{y}} \delta u\right) \ dxdy \end{equation*}$ which then equals ... (second comment) $\endgroup$ – user1887239 Dec 30 '18 at 13:17
  • $\begingroup$ $ \begin{equation*} \oint _{s}\left(\left(\frac{\partial F}{\partial u^{'}_{x}}\right) \ \overrightarrow{n_{x}} +\ \left(\frac{\partial G}{\partial u^{'}_{y}}\right)\overrightarrow{n_{y}} \ \right) \delta u\ ds \end{equation*}$ (n being the unit vector) I don't know how to apply the same thing here $\endgroup$ – user1887239 Dec 30 '18 at 13:17
  • $\begingroup$ But you now have an integral equation over a rectangle, its exactly the same. Just substitute $r=x$, $\theta=y$ and see how far you can get forgetting about polar coordinates. $\endgroup$ – Eddy Dec 30 '18 at 13:20

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.