Mathematics Stack Exchange is a question and answer site for people studying math at any level and professionals in related fields. Join them; it only takes a minute:

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

given the 2 PDE

$$ \Delta u-au_{tt}+u_{t}=0$$

and $$ \Delta u + Du*Df=0 $$

here $ \Delta u $ is the Laplacian $ Du= grad(u) $ is the gradient and * means scalar product $u_{t} = \frac{\partial u}{\partial t}$

my doubt is what term should i include to get the linear part of the equations $ u_{t} $ and $Du*Df $ from the Euler Lagrange equations in $ R^{n} $ thanks

share|cite|improve this question

migrated from Feb 22 '14 at 3:29

This question came from our site for professional mathematicians.

up vote 2 down vote accepted

Well, for the second case, the Lagrangian you want is $$ \Lambda_D(u) = \tfrac12 \int_D\ e^{f(x)}\ |\nabla u|^2\ \mathrm{d}x $$ where $D$ is an appropriate domain in $\mathbb{R}^n$. The first case is similar (if you are assuming that $a$ is a constant), for then, it will just be $$ \Lambda_D(u) = \tfrac12 \int_{D'}\ e^{t}\ \bigl(|\nabla u|^2-a|u_t|^2\bigr)\ \mathrm{d}x\mathrm{d}t $$ where $D'$ is an appropriate domain in $\mathbb{R}^n\times \mathbb{R}$. If $a$ is not a constant, there are conditions.

share|cite|improve this answer
thank you RObert :D wehn i have enoguht poitn to vote i will give you a point thanks again – Jose Garcia Feb 18 '14 at 20:54
shouldn't this be the term $ e^{f(x)}$ instead ?= in both cases ?? – Jose Garcia Feb 18 '14 at 20:57
Yes, you are right. I fixed the sign a few moments ago. – Robert Bryant Feb 18 '14 at 20:58

Here we will assume that the definition of the Laplacian is $\Delta=\vec{\nabla}^2$, and that $a$ is a non-zero constant.

I) The Lagrangian density

$$\tag{1a} {\cal L}_1~=~ \frac{e^{-t/a}}{2}\left(a\dot{u}^2 - (\vec{\nabla} u)^2\right)$$

has Euler-Lagrange equation

$$0~\stackrel{(1a)}{=}~\frac{\partial {\cal L}_1}{\partial u} ~\stackrel{EL}{=}~\frac{d}{dt}\frac{\partial {\cal L}_1}{\partial \dot{u}} +\vec{\nabla}\cdot\frac{\partial {\cal L}_1}{\partial \vec{\nabla}u}$$ $$\tag{1b} ~\stackrel{(1a)}{=}~ \frac{d}{dt}\left(e^{-t/a}a\dot{u}\right) - \vec{\nabla}\cdot \left(e^{-t/a}\vec{\nabla}u\right) ~=~e^{-t/a}\left(a\ddot{u}-\dot{u}-\Delta u\right) $$

equal to OP's first PDE

$$\tag{1c} a\ddot{u}-\dot{u}-\Delta u~\stackrel{(1b)}{=}~0.$$

II) The Lagrangian density

$$\tag{2a} {\cal L}_2~=~ \frac{e^f}{2} (\vec{\nabla} u)^2$$

has Euler-Lagrange equation

$$ 0~\stackrel{(2a)}{=}~\frac{\partial {\cal L}_2}{\partial u} ~\stackrel{EL}{=}~\vec{\nabla}\cdot\frac{\partial {\cal L}_2}{\partial \vec{\nabla}u}$$ $$\tag{2b}~\stackrel{(2a)}{=}~\vec{\nabla}\cdot \left(e^f\vec{\nabla}u\right) ~=~e^f\left(\Delta u +\vec{\nabla}f \cdot \vec{\nabla}u\right) $$

equal to OP's second PDE

$$\tag{2c} \Delta u +\vec{\nabla}f \cdot \vec{\nabla}u~\stackrel{(2b)}{=}~0.$$


[Note added: OP's question was originally cross-posted to mathoverflow and migrated back to Math.SE, so that OP's question appeared in duplicates on Math.SE. This answer (which was composed without knowing of the duplicate entry) originally appeared under the original Math.SE question, but is now moved here to collect all answers in one place.]

share|cite|improve this answer
THANKS a lot could yuou help me with other problems i have about variatonal calculus :D i must prove 20 exercises but i have only solved 4 :D – Jose Garcia Mar 15 '14 at 20:44

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.