I want to show that if $\phi$ transforms as the adjoint representation of SU(n), its covariant derivative is given by $\textbf{D}_\mu \phi = \partial_\mu \phi + i [\textbf{A}_\mu, \phi]$. (Exercise in chapter 6 of P. Ramond QFT book)

The covariant derivative is defined as

$$\textbf{D}_\mu= \partial_\mu + i\textbf{A}_\mu$$

If $\phi$ have the transformation properties assumed, we know that $\phi= \phi^a T^a$, where $T^a$ are the generators of SU(n) and in the adjoint representation $(T^b)_{cd}= -if^{bcd} $. Besides that, we have $\textbf{A}_\mu=A_\mu^a T^a$, etc.

The attempted solution:

$$\begin{split}(\textbf{D}_\mu \phi)_{ab} & = (\partial_\mu \phi)_{ab} + i (\textbf{A}_\mu \phi)_{ab} \\ & = (\partial_\mu \phi)_{ab} + i A_\mu^c \phi^e (T^c)_{ad}(T^e)_{db} \end{split}$$

And now I can use the relation $(T^b)_{cd}= -if^{bcd} $, but it doesn't get me to the right answer because there isn't one $f$ that contracts with both $A$ and $\phi$ as that should be to get the correct result. We can see this writing the answer with explicit components

$$(\textbf{D}_\mu \phi)_{ab} = (\partial_\mu \phi)_{ab} - A_\mu^d \phi^e f^{dec}(T^c)_{ab}.$$

How should I proceed? Any help will be appreciated.


Following the comments, I got to a point and now I am stuck:

$$\begin{split}(\textbf{D}_\mu \phi)_{be} & = (\partial_\mu \phi)_{be} + i A_\mu^a \phi^d (T^a)_{bc}(T^d)_{ce} \\ & = (\partial_\mu \phi)_{be} + i A_\mu^a \phi^d (-if^{abc})(-if^{dce}) \\ & = (\partial_\mu \phi)_{be} + i A_\mu^a \phi^d (-f^{adc}f^{ceb}-f^{aec}f^{cbd}) \\ & = (\partial_\mu \phi)_{be} - A_\mu^a \phi^d f^{adc}(T^c)_{be}-A_\mu^a \phi^d f^{aec}f^{cbd} \\ & =(\partial_\mu \phi)_{be} +i A_\mu^a \phi^d [T^a,T^d]_{be}-A_\mu^a \phi^d f^{aec}f^{cbd}\end{split} $$

Which shows that I got the right answer plus $-A_\mu^a \phi^d f^{aec}f^{cbd}$. Is this right so far?

  • $\begingroup$ But am I not supposed to start from the definition $\textbf{D}_\mu \phi = (\partial_\mu + i \textbf{A}_\mu ) \phi$ an then manipulate assuming $\phi = \phi^a T^a$ to get to $\textbf{D}_\mu \phi = \partial_\mu \phi + i[\textbf{A}\mu, \phi]$? $\endgroup$ – Slayer147 Apr 10 at 11:55
  • $\begingroup$ Did you fill in, the attempted solution, $\left( T^e\right )_db=-if^{edb}$? $\endgroup$ – descheleschilder Apr 10 at 13:27
  • $\begingroup$ Yes, and it is done in my edit too. $\endgroup$ – Slayer147 Apr 10 at 13:39

I see now the problem. The definition of the covariant derivative in a general representation $r$ for a field $\phi_\alpha$, with $\alpha = 1,\ldots \mathrm{dim}\,r$, is $$ (\mathbf{D}_\mu\phi)_\alpha = \partial_\mu\phi_\alpha + i A_\mu^a (T^a_r)_{\alpha\beta}\phi_\beta\,. $$ Where $T^a_r$ are generators in the representation $r$.

A field in the adjoint representation of an algebra $\mathfrak{g}$ can be seen as a vector with an index $a = 1,\ldots,\mathrm{dim}\,\mathfrak{g}$, or as a matrix with two indices belonging to any representation. If we denote it as a matrix the definition of the covariant derivative is via the commutator.

This is what the exercise asks for:

Show that these two definitions of the covariant derivative in the adjoint representation are equivalent: $$ (\mathbf{D}_\mu\phi)_{a} = \partial_\mu\phi_a + i A_\mu^c (T^c_\mathfrak{g})_{ab}\phi_b\,, $$ $$ (\mathbf{D}_\mu\phi)_{\alpha\beta} = \partial_\mu\phi_{\alpha\beta} + i [\mathbf{A}_\mu, \phi]_{\alpha\beta}\,. $$

The indices in the second equation can be any representation. As you correctly stated $$ (T_\mathfrak{g}^c)_{ab} = -i f^{cab}\,. $$ In order to show the claim one needs to multiply the first equation by $T^a_r$.

$$ \partial_\mu\phi_{\alpha\beta} + A_\mu^c (T^a_r)_{\alpha\beta}f^{cab}\phi_b = \partial_\mu \phi_{\alpha\beta} + i A_\mu^c [T^c_r,T^b_r]\phi^b\,. $$

This is true and one does not even need the Jacobi identity. If you choose $r$ to be the adjoint, this does need the Jacobi, but only because in the adjoint representation the Jacobi plays the role of the commutation rule.

  • $\begingroup$ Thanks for the answer. I just don't understand your claim that "if you choose $r$ to be the adjoint, this does need the Jecobi". From what I see I can get the same answer doing the same steps above even if $r$ is the adjoint. $\endgroup$ – Slayer147 Apr 10 at 18:17
  • 1
    $\begingroup$ Probably I shouldn't have said that. What I meant is that the statement $[T^a,T^b]= i f^{abc} T^c$ is the Jacobi identity if expressed in the adjoint representation. $\endgroup$ – MannyC Apr 10 at 18:27

Remember that the structure constants satisfy the Jacobi-relation

$$ f^{abc}f^{cde} + f^{adc}f^{ceb} + f^{aec}f^{cbd} = 0$$

and that the gauge field also transforms under the adjoint representation. Using this, i think it should be straightforward to prove the statement.

  • $\begingroup$ I edited the question. Following your comment, I got to the point above. $\endgroup$ – Slayer147 Apr 10 at 2:00

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.