Physics Stack Exchange is a question and answer site for active researchers, academics and students of physics. 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

I'm trying to understand the proof of Noether's Theorem in my Classical Mechanics class. We formulated it as follows:

A continous symmetry is defined as a flow $\phi^{\lambda}(q(t))$ which leaves the Lagrangian invariant, i.e.

$$L(\phi^{\lambda}(q(t)), \frac{\partial}{\partial t}\phi^{\lambda}(q(t)),t) = L(q(t),\dot{q}(t),t)$$

for each $\lambda$ and each $q(t)$ which solves the Euler-Lagrange equations.

The conserved quantity then is

$$\langle p,v(q)\rangle = \sum_{\alpha =1}^f p_{\alpha} v^{\alpha}(q)$$

where $p_{\alpha} = \frac{\partial L}{\partial \dot{q}^{\alpha}}$ is the conjugated momentum to $q^{\alpha}$ and $v(q)$ is the vector field implied by the flow $\phi^{\lambda}(q(t))$.

So the proof goes as follows:

$$\begin{align} \frac{\mathrm{d}}{\mathrm{d}t} \Biggl[\sum_{\alpha =1}^f p_{\alpha} v^{\alpha}(q)\Biggr] &= \sum_{\alpha =1}^f \biggl[\frac{\mathrm{d}}{\mathrm{d}t}\frac{\partial L}{\partial \dot{q}^{\alpha}} \biggr] v^{\alpha}(q) + \sum_{\alpha =1}^f \frac{\partial L}{\partial \dot{q}^{\alpha}} \frac{\partial}{\partial t}v^{\alpha}(q) \\ &= \sum_{\alpha =1}^f \frac{\partial L}{\partial q^{\alpha}} \frac{\partial}{\partial \lambda} \phi^{\lambda}\bigl(q(t)\bigr) + \sum_{\alpha =1}^f \frac{\partial L}{\partial \dot{q}^{\alpha}} \frac{\partial}{\partial \lambda}\frac{\partial}{\partial t}\phi^{\lambda}\bigl(q(t)\bigr) \\ &= \frac{\mathrm{d}}{\mathrm{d}\lambda} L\biggl(\phi^{\lambda}\bigl(q(t)\bigr), \frac{\partial}{\partial t} \phi^{\lambda}\bigl(q(t)\bigr),t\biggr) \\ &= 0 \end{align}$$

where we used the Euler-Lagrange equations in the second line and the definition of the vector field $v(q)$. In the third line we used the definition of a continuous symmetry.

There's just one small thing that I don't fully understand now: If we calculate

$$\frac{\mathrm{d}}{\mathrm{d}\lambda} L\biggl(\phi^{\lambda}\bigl(q(t)\bigr), \frac{\partial}{\partial t} \phi^{\lambda}\bigl(q(t)\bigr),t\biggr) = \frac{\partial L}{\partial \phi^{\lambda}}\frac{\partial}{\partial \lambda} \phi^{\lambda} + \frac{\partial L}{\partial (\frac{\partial}{\partial t} \phi^{\lambda} )} \frac{\partial}{\partial \lambda} \frac{\partial }{\partial t} \phi^{\lambda}$$

(I dropped the summation here). Why is the derivative towards $\phi^{\lambda}$ immediately the same as the derivative towards $q(t)$? I know the flow $\phi^{\lambda}$ maps physical solutions to physical solutions but I don't quite intuitively see how the two derivatives are the same. I hope someone can clear this up for me!

share|cite|improve this question
in both cases, it's the partial derivative of $L$ with respect to its first variable; it might be less confusing if you'd given the variable and the trajectory a different name, eg $L=L(x,v,t)$ and use $\partial L/\partial x$, regardless if you evaluate the function at $(q(t),\dot q(t), t)$ or $(\phi^\lambda(q(t)),\partial_t\phi^\lambda(q(t)),t)$ – Christoph Jun 9 '13 at 10:29
@Christoph But why does choosing a different variable for the trajectory that depends on the old variable not change the function? If $\phi^{\lambda}$ where no function of $q(t)$ I'd see what you mean, but if we express our Langrangian through a new variable which is a function of the old variables, the Lagrangian should change, and hence the derivative should too. What am I missing? – user17574 Jun 9 '13 at 10:46
it's the chain rule - regardless if you're looking at $L$, $L\circ q$ or $L\circ\phi^\lambda\circ q$, the leftmost term will always be $\partial L/\partial x$ – Christoph Jun 9 '13 at 12:04

Your Answer


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

Browse other questions tagged or ask your own question.