Knowing that the free Dirac Lagrangian is :

$$\tag{1} \mathcal{L}= \bar{\psi} (i \gamma^\mu \partial_\mu -m ) \psi$$

and that the Euler-Lagrange equation is:

$$\tag{2} \frac{\partial \mathcal{L}}{\partial \psi}= \partial_\mu \left( \frac{\partial \mathcal{L}}{\partial(\partial_\mu \psi)}\right)$$

I am trying to obtain the standard form of the Dirac equation $(3)$ without differentiating by $\bar{\psi}$:

$$\tag{3} (i\gamma^\mu \partial_\mu - m) \psi =0$$

I've been told that the relations $(\gamma^0)^2=1$ and $\gamma^{\dagger \mu}= \gamma^0 \gamma^\mu \gamma^0$ might be useful (I don't see how)

My take on it:

After expanding $(1)$: $$\tag{4} \frac{\partial \mathcal{L}}{\partial \psi}=- \bar{\psi}m$$


$$\tag{5} \frac{\partial \mathcal{L}}{\partial ( \partial_\mu \psi)}=\bar{\psi} i \gamma^\mu$$


$$\tag{6} -\bar{\psi}m - \partial_\mu (\bar{\psi}i \gamma^\mu)=0$$

this is where I become lost:

I think I can use the first term on $(6)$ and rewrite it as $ -\bar{\psi}m= m \bar{\psi}$, but how do I proceed from here?

I think I must find a way to place on $\bar{\psi}$ on the RHS of both terms in $(6)$, proceed by multiplying everything by $\psi$ twice, once to cancel the $\bar{\psi}$ and twice to terminate in the standard form shown in $(3)$. How do I do this? Is there another way?

I have seen other questions and links on the website but these don't quite do it as I intend and tend to differentiate by $\bar{\psi}$ which isn't what I want. Some of the links visited are: Going from the Dirac Lagrangian to the adjoint Dirac equation , Derivation of Dirac equation using the Lagrangian density for Dirac field .

  • 1
    $\begingroup$ Hint: Derive the Euler-Lagrange-Equation of $\bar\psi$. $\endgroup$
    – Toffomat
    Jun 2, 2020 at 11:00
  • $\begingroup$ That has already been done (in one of the links above it is done that way) but I am trying to do it by differentiating via $\psi$ and not $\bar{\psi}$. Thank you for your help $\endgroup$ Jun 2, 2020 at 11:02
  • $\begingroup$ Why did you consider $\psi$ and $\bar{\psi}$ as independent variables? $\endgroup$
    – Nikita
    Jun 30, 2020 at 10:27

2 Answers 2


You start with $$\bar{\psi}(i\gamma^\mu\overleftarrow{\partial}_\mu+m) = 0$$ if you take the hermitian conjugate of all

$$\begin{align}\left(\bar{\psi}(i\gamma^\mu\overleftarrow{\partial}_\mu+m)\right)^\dagger &= \left(\psi^\dagger\gamma^0(i\gamma^\mu\overleftarrow{\partial}_\mu+m)\right)^\dagger= (i\gamma^\mu\overleftarrow{\partial}_\mu+m)^\dagger\gamma^{0\dagger}\psi\\ &= (-i\gamma^{\mu\dagger}\partial_\mu+m)\gamma^0\psi = (-i\gamma^0\gamma^\mu\gamma^0\partial_\mu+m)\gamma^0\psi\\ &=(-i\gamma^0\gamma^\mu\gamma^0\gamma^0\partial_\mu+m\gamma^0)\psi =(-i\gamma^0\gamma^\mu\partial_\mu+\gamma^0m)\psi\\ &=\gamma^0(-i\gamma^\mu\partial_\mu+m)\psi= 0 \end{align}$$ Now multiply from the left with a $\gamma^0$ and use the property $\gamma^0\gamma^0=1$ $$(-i\gamma^\mu\partial_\mu+m)\psi=0$$ which is exactily $$(i\gamma^\mu\partial_\mu-m)\psi = 0$$

  • $\begingroup$ I understand that in the first line you can move the $\gamma^0$ outside of the differentiation because it is just a number, but how can you just move the $\psi^\dagger$ by multiplying it by $\psi$? I thought we couldn't do it, because it is being differentiated. $\endgroup$ Jun 2, 2020 at 13:27
  • $\begingroup$ Nobody said that we're multiplying by $\psi$ since by doing so you won't achieve anything at all. $\psi$ is a spinor, a function, it makes no sense multiply something by it. In the first line we only used the fact that transposition swaps the terms in a product $(AB)^T = B^T A^T$ $\endgroup$ Jun 2, 2020 at 14:48

Your question is based on a false premise. You can't get it except by differentiating with respect to $\bar{ψ}$. If you differentiate with respect to $ψ$, you get the adjoint Dirac equation for $\bar{ψ}$.

If you differentiate with respect to $ψ$, to get the adjoint equation for $\bar{ψ}$, and then take the adjoint of the equation - then all you've really done was just differentiate with respect to $\bar{ψ}$ in the first place. So, no matter what you do, it all comes back to that.

You have to treat $ψ$ and $\bar{ψ}$ as independent variables. So, you're getting two equations, not one. You can see that more clearly in my reply here. In fact, as I point out there, it's actually more correct to write the Lagrangian density symmetrically as: $$\mathcal{L} = \bar{ψ} \left(\frac{iħ}{2} \left(γ^μ \overrightarrow{∂_μ} - \overleftarrow{∂_μ} γ^μ\right) - mc\right) ψ,$$ as some authors do. Then, in the variational: $$Δ\mathcal{L} = Δ\bar{ψ} \left(iħ γ^μ \overrightarrow{∂_μ} - mc\right) ψ - \bar{ψ}\left(iħ \overleftarrow{∂_μ} γ^μ + mc\right) Δψ + ∂_μ \left(\frac{iħ}{2} \bar{ψ} \left(γ^μ \overrightarrow{Δ} - \overleftarrow{Δ} γ^μ\right) ψ\right),$$ you get symmetric contributions from $\overrightarrow{Δ}ψ ≡ Δψ$ and $\bar{ψ}\overleftarrow{Δ} ≡ Δ\bar{ψ}$ to the boundary term.


Your Answer

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