I'm trying to compute a result shown in my classical mechanics lecture on my own. Namely, consider that a system composed of $n$ particles follows a law of force

$m_k\ddot{\vec{x_k}} = \vec{F_k}(\vec{x_i};\dot{\vec{x_i}};t)$, (1)

where we write $i$, but we are actually considering all positions and velocities of the system as arguments of $F_k$. Now, if the force can be written as a time-independent potential

$F_k = \frac{\partial}{\partial \vec{x_k}}V(\vec{x_1},\dots, \vec{x_n})$, (2)

we can write the Galilei invariance characterised by a rotation $R$ and a translation $\vec{b}$ as follows:

$R\frac{\partial}{\partial \vec{x_k}}V(\vec{x_1},\dots, \vec{x_n}) = \frac{\partial}{\partial \vec{x_k'}}V(\vec{x_1'},\dots, \vec{x_n'})$, (3)

where $\vec{x_i'} = R\vec{x_i} + \vec{b}$ is the Galilei-Transformation of $\vec{x_i}$. In my lecture, it was stated that because of this we can write:

$V(\vec{x_1},\dots, \vec{x_n}) = V(\vec{x_1'},\dots, \vec{x_n'})$. (4)

Now, I'm struggling to prove why (4) follows from (3). I tried using a naive chain rule on the inverse Galilei-transformation $\vec{x_k'} \mapsto \vec{x_k}$ but I always end up with:

$R\frac{\partial}{\partial \vec{x_k}}V(\vec{x_1},\dots, \vec{x_n}) = R^T\frac{\partial}{\partial \vec{x_k}}V(\vec{x_1'},\dots, \vec{x_n'})$. (5)

And I haven't yet found a way to imply (3) using (5). Now, I'm not sure if my calculations are right, or if I have missed something to prove that (3) follows from (5). I'd rather say that I've been misusing the chain rule.


1 Answer 1


The potential $V$ is a scalar quantity, so it is invariant under coordinate transformations. As for why 3 implies 4, I think you have to be more careful about how you're defining your gradient.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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