Suppose we have the Lagrangian density for a triplet of real scalar fields, $$ L = \sum_{a=1}^3 \left[ \frac{1}{2}\partial_\mu\phi_a\partial^\mu\phi_a - \frac{1}{2}\phi_a\phi_a \right]. $$

How do you show that it's invariant under the infinitesimal $SO(3)$ transformation $$ \phi_a \to \phi_a + \theta \epsilon_{abc}n_b\phi_c, $$

where $\theta$ is a constant and $n_b$ a unit vector. I'm not really sure actually what to do after simply substituting $\phi_a$.


1 Answer 1


This is easiest done not using index notation. Define the column vector

$$ \vec{\phi} = \begin{pmatrix} \phi_1 \\ \phi_2 \\ \phi_3 \end{pmatrix} $$

The derivative acts on the column vector as $\mathbf{1}_3\partial_\mu$ where $\mathbf{1}_3$ is the identity matrix on the same vector space in which $\vec{\phi}$ lives. That is,

$$ \partial_\mu \vec{\phi} = \begin{pmatrix} \partial_\mu \phi_1 \\ \partial_\mu \phi_2 \\ \partial_\mu \phi_3 \end{pmatrix} $$

Then we can write the lagrangian as

$$ \mathcal{L}=\frac{1}{2}(\partial_\mu \vec{\phi})^T (\partial_\mu \vec{\phi}) - \frac{1}{2}\vec{\phi}^T \vec{\phi} $$

Now it can be seen basically by inspection that for $R\in SO(3)$ we have that

$$ \vec{\phi} \stackrel{SO(3)}{\longmapsto} R\vec{\phi} \tag{(a)}\\ \vec{\phi}^T \stackrel{SO(3)}{\longmapsto} \vec{\phi}^T R^T $$

Under this transfermation, both terms in the lagrangian both terms pick up an $R^TR = \mathbf{1}_3$ sandwiched in the middle. Hence the lagrangian is invariant.

Relating to the transformation law using index notation:

$$ \phi_a \to \phi_a + \theta \epsilon_{abc}n_b\phi_c $$

A general $SO(3)$ transformation can be written as

$$ R = e^{-iL_an_a \theta} $$

so that an infinitesimal transformation can be written as

$$ R = 1 -iL_an_a \theta + \mathcal{O}(\theta^2) $$

Now, in the fundamental representation the generators are given by

$$ (L^a)_{bc} = -\epsilon_{abc} $$

We therefore see that an infinitesimal SO(3) transformation on $\phi_a$ is given as

$$ \phi_a \to R_{ab}\phi_b = \phi_a + i\theta\epsilon_{abc}n_c \phi_b + \mathcal{O}(\theta^2) \tag{b} $$

So we see that proving the lagrangian is invariant under (b) is a special case of proving it is invariant under (a).

  • $\begingroup$ Thank you, but how would you relate this to the $+\theta \epsilon_{abc}n_b\phi_c$ part. It seems as if you ignored that, no? $\endgroup$ Nov 12, 2019 at 17:39
  • 1
    $\begingroup$ I'll add an edit. In short they're identical, but you can show it for any SO(3) matrix, not just the infinitesimal transformations. $\endgroup$ Nov 12, 2019 at 17:40
  • $\begingroup$ While this is correct, I think it misses the point of the exercise, which is presumably to build up lie groups from their generators. Edit: your edit makes this comment irrelevant. $\endgroup$ Nov 12, 2019 at 17:55

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.