I roughly understand the concept of the Lagrangian $L = T - V$ as well as the idea of stationary action $\delta \mathcal{S} =0$. However, I am confused what the Euler-Lagrange equation actually says.

Consider the Euler-Lagrange equation: \begin{equation} \frac{\partial L}{\partial q} - \frac{d}{dt} \left(\frac{\partial L}{\partial \dot{q}}\right) = 0 \end{equation}

Here's my confusion:

To me, this looks like an empty mathematical exercise. If I know the partial derivatives $\frac{\partial L}{\partial q}$ and $\frac{\partial L}{\partial \dot{q}}$ and can take the derivative with respect to $t$ of the latter, what is the use of plugging all that into this setup? That's like saying after I show $2 + 3 = 5$, then show $x + y = z$, where $x=2,y=3,z=5$. In short, an empty exercise since it would be the same proof showing $x + y = z$ as showing $2 +3 = 5$ in this case.

Can someone explain what I am misunderstanding?

  • $\begingroup$ I'm not sure I understand what you're asking. Given a Lagrangian, the E-L equations tell you what the equations of motion are. $\endgroup$
    – Jold
    Jan 22 '15 at 5:11
  • $\begingroup$ Yeah, I think that's what I'm not grasping. For instance, consider dropping an object of mass $m$ from a height $h$ off the surface of the earth. $mgh$ is PE $\frac{1}{2}mv^2$ is KE. What does E-L tell me about the path that I don't already know given a vector field model of gravity. I know the path is towards the earths surface already. What can E-L tell me? $\endgroup$ Jan 22 '15 at 5:14
  • $\begingroup$ Here is a good introduction. damtp.cam.ac.uk/user/tong/dynamics.html $\endgroup$
    – mmesser314
    Jan 22 '15 at 5:28

The reason algebra ($x + y = z$) is more useful than arithmetic ($2 + 3 = 5$) is that it allows you to treat large classes of problems in the same way, and to obtain very general results about them without having to do a separate calculation for every individual case. Same goes for the stationary action principle (a.k.a. action minimization). It allows you to treat a large class of problems - in this case, all of physics - in the same way, by solving $\delta S = 0$. Different choices of $S$ correspond to different areas of physics, but once you have $S$, the procedure for solving is always the same, namely the Euler-Lagrange equations.

Perhaps you've only been exposed to one formula for the action, $S = \int (T - V)\,\mathrm{d}t$ (or equivalently $L = T - V$), which is the action for a Newtonian system, and in that case it's understandable that the stationary action principle seems kind of useless. Similarly, if you had only been exposed to the one example of $x + y = z$ where $x = 2$, $y = 3$, and $z = 5$, it would seem pretty pointless to use letters. But it makes more sense once you see how general the formalism is.

  • $\begingroup$ Oh, that is probably exactly the problem. I learned about applications in my geodesic exercises in general relativity today and for the first time I started to see how it could be more powerful than just the Newtonian case, as you say. $\endgroup$ Jan 22 '15 at 5:57
  • $\begingroup$ What are some fundamental example uses for the Euler-Lagrange equation that go beyond the Newtonian cases so I can start to expand my understanding? I assume any elementary text will cover them, but will help if I know which examples to pay particular attention to. $\endgroup$ Jan 22 '15 at 5:59
  • $\begingroup$ Any area of physics you want to explore, there's probably an action for it. For example general relativity uses the Einstein-Hilbert action. In this case the Euler-Lagrange equations give you the Einstein field equations. There are also actions for electromagnetism, relativistic particles, quantum field theory, etc. $\endgroup$
    – David Z
    Jan 22 '15 at 7:36

Euler-Lagrange equations are a necessary condition if the action integral reaches an extremal through the trajectory described by $q(t)$ and you should obtain an expression for $q(t)$ for a given Lagrangian $L$ by solving Euler-Lagrange equations. Take for example the simple Lagrangian for a free particle with constant mass: $L=m\dot{q}^2/2$ then E-L equations are

\begin{equation} \frac{d}{dt}\left(m\dot{q}\right)=m\ddot{q}=0 \end{equation}

which has for solution

\begin{equation} q(t)=at+b \end{equation}

a straight line, something you already knew from Newton first law of motion.

Try solving E-L equations with the Lagrangian $L=m\dot{z}^2/2-mgz$ for the particle in free fall (taking partial derivatives with respect to $z$ and $\dot{z}$ instead of $q$ and $\dot{q}$) to obtain the well known trajectory $z(t)=gt^2/2+\dot{z}_0t+z_0$, where $z_0$ and $\dot{z}_0$ are the initial position and velocity.

  • $\begingroup$ Ohh, wait. Okay, maybe I am starting to understand. So are they merely a necessary condition or are they a sufficient condition to tell us if the action integral reaches an extremal through the trajectory described by $q(t)$? $\endgroup$ Jan 22 '15 at 5:37
  • $\begingroup$ They are not a sufficient condition, but i can't remember of a counterexample right now. To derive the E-L equations you assume that there exist a trajectory which minimizes the action and then you find out that this trajectory must satisfy the E-L equations. In fact there are several ways to derive these equations, look at the book "Emmy Noether's Wonderful Theorem" the author is Dwight E. Neuenschwander, there you'll find a good discussion about the equivalence of the principle of minimum action and Newton's second law. $\endgroup$
    – C. Esparza
    Jan 22 '15 at 7:40

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.