A lagrangian $L(q,\dot q, t)$ is invariant under the point transformation $$q_i=q_i(s_1,...,s_n,t)$$

To prove this I show that

$$\frac{d}{dt} \frac{\partial L}{\partial \dot s_i} - \frac{\partial L}{\partial s_i} = 0$$


$$\frac{\partial L}{\partial s_i} = \frac{\partial L}{\partial q_j}\frac{\partial q_j}{\partial s_i}$$

$$\frac{\partial L}{\partial \dot s_i} = \frac{\partial L}{\partial \dot q_j}\frac{\partial \dot q_j}{\partial \dot s_i}$$

$$\frac{\partial \dot q_j}{\partial \dot s_i}= \frac{\partial q_j}{\partial s_i}$$

which gives $$\frac{d}{dt} \left(\frac{\partial L}{\partial \dot q_j}\frac{\partial q_j}{\partial s_i} \right) - \frac{\partial L}{\partial q_j}\frac{\partial q_j}{\partial s_i} = 0$$

but in order to get into the form $$\left(\frac{d}{dt} \frac{\partial L}{\partial \dot q_j} - \frac{\partial L}{\partial q_j}\right)\frac{\partial q_j}{\partial s_i} = 0$$

I need to prove that $$\frac{d}{dt} \left(\frac{\partial L}{\partial \dot q_j}\frac{\partial q_j}{\partial s_i} \right) = \frac{d}{dt} \left(\frac{\partial L}{\partial \dot q_j}\right)\frac{\partial q_j}{\partial s_i} $$

What is the logic behind this? Any help would be greatly appreciated!

  • 1
    $\begingroup$ This equation is not correct: $$\frac{\partial L}{\partial s_i} = \frac{\partial L}{\partial q_j}\frac{\partial q_j}{\partial s_i}$$ It should be $$ \frac{\partial L}{\partial s_i} = \frac{\partial L}{\partial q_j}\frac{\partial q_j}{\partial s_i} + \frac{\partial L}{\partial \dot q_j} \frac{\partial \dot q_j}{\partial s_i} $$ $\endgroup$
    – md2perpe
    Jul 23, 2018 at 20:03

1 Answer 1


In case anyone wants the whole proof.

Invariance of Lagrange's Equations under Point Transformations

Let $q_1,q_2,\cdots,q_n$ be a set of independent generalised coordinates for a system on $n$ degrees of freedom, with Lagrangian $L(q,\dot{q},t)$. Suppose we transform to another set of independent coordinates $s=s_1,s_2,\cdots,s_n$ be means of transformation equations: \begin{equation} q_i=q_i(s,t) \qquad i=1,2,\cdots,n \qquad (1) \end{equation} and \begin{align} \dot{q}_i=\frac{dq_i}{dt}=\sum_j \frac{\partial q_i}{\partial s_j}\dot{s}_j+\frac{\partial q_i}{\partial t} \nonumber \\ \dot{q}_i=\dot{q_i}(s,\dot{s},t)\qquad i=1,\cdots,n \qquad (2) \end{align} Such a transformation is called a point transformation. Then: \begin{align} \frac{\partial L}{\partial s_j}=\sum_i\left[\frac{\partial L}{\partial q_i}\frac{\partial q_i}{\partial s_j} +\frac{\partial L}{\partial \dot{q}_i}\frac{\partial \dot{q}_i}{\partial s_j}\right]\qquad (3) \end{align}

\begin{align} \frac{d}{dt}\left(\frac{\partial L}{\partial \dot{s}_j}\right)&=\frac{d}{dt}\left[\sum_i\left(\frac{\partial L}{\partial q_i}\frac{\partial q_i}{\partial \dot{s}_j}+\frac{\partial L}{\partial \dot{q}_i}\frac{\partial \dot{q}_i}{\partial \dot{s}_j}\right) \right] \qquad (4) \end{align} but we also know that $\partial q_i/\partial \dot{s}_j=0$ and also: \begin{align} dq_i=\sum_j\frac{\partial q_i}{\partial s_j}ds_j+\frac{\partial q_i}{\partial t}dt \nonumber \\ \frac{dq_i}{dt}=\sum_j\frac{\partial q_i}{\partial s_j}\frac{ds_j}{dt}+\frac{\partial q_i}{\partial t} \nonumber \\ \dot{q}_i=\sum_j\frac{\partial q_i}{\partial s_j}\dot{s}_j+\frac{\partial q_i}{\partial t} \nonumber \end{align} \begin{equation} \frac{\partial \dot{q}_i}{\partial \dot{s}_j}=\frac{\partial q_i}{\partial s_j} \qquad (5) \end{equation} Making these changes to (4) we get: \begin{align} \frac{d}{dt}\left(\frac{\partial L}{\partial \dot{s}_j}\right)&=\frac{d}{dt}\left[\sum_i \frac{\partial L}{\partial \dot{q}_i}\frac{\partial q_i}{\partial s_j}\right] =\nonumber \\ &=\sum_i \left[\frac{d}{dt}\left(\frac{\partial L}{\partial \dot{q}_i}\right)\frac{\partial q_i}{\partial s_j}+\frac{\partial L}{\partial \dot{q}_i}\frac{d}{dt}\left(\frac{\partial q_i}{\partial s_j}\right)\right] \qquad (6) \end{align} But we know that $L$ satisfies the Lagrangian equations for the $q$ set of coordinates. Hence: \begin{equation} \frac{d}{dt}\left(\frac{\partial L}{\partial \dot{q}_i}\right)=\frac{\partial L}{\partial q_i} \qquad (7) \end{equation} as well as the property: \begin{align} \frac{d}{dt}\left(\frac{\partial q_i}{\partial s_j}\right)&=\sum_k\frac{\partial}{\partial q_k}\left(\frac{\partial q_i}{\partial s_j}\right)\dot{q_k}+\frac{\partial^2q_i}{\partial t\partial s_j}=\nonumber \\ &=\frac{\partial}{\partial s_j}\left[\sum_k\frac{\partial q_i}{\partial q_k}\dot{q}_k+\frac{\partial q_i}{\partial t}\right]\nonumber \\ &=\frac{\partial}{\partial s_j}\left(\frac{d q_i}{dt}\right)=\nonumber \\ &=\frac{\partial \dot{q}_i}{\partial s_j}\qquad (8) \end{align} and (6) becomes \begin{align} \frac{d}{dt}\left(\frac{\partial L}{\partial \dot{s}_j}\right)&=\sum_i\left[\frac{\partial L}{\partial q_i}\frac{\partial q_i}{\partial s_j}+\frac{\partial L}{\partial \dot{q}_i}\frac{\partial \dot{q}_i}{\partial s_j}\right] \qquad(9) \end{align}

Now making the substitutions (3) and (9), we get the important result that: \begin{align} \frac{d}{dt}\left(\frac{\partial L}{\partial \dot{s}_j}\right)-\frac{\partial L}{\partial s_j}=0 \qquad(10) \end{align}

  • $\begingroup$ Hi, nice write up! I have a question about equations (3) and (4). Why we don't have to include the terms involving $\frac{\partial L}{\partial t}$, while in the equation just below (4), you have the term $\frac{\partial q_i}{\partial t} dt$? $\endgroup$
    – hb12ah
    Apr 2, 2020 at 7:23
  • $\begingroup$ I think I've asked a bad question. Now I see the reason. $q$ is a function of $s$ but $t$ is not. $\endgroup$
    – hb12ah
    Apr 2, 2020 at 9:26
  • $\begingroup$ Nice answer, just a typo in the third line of (8), should be $\frac{dq_i}{dt}$ instead of $\frac{dq}{dt}$ $\endgroup$
    – ydhhat
    Oct 24, 2020 at 16:22

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.