There is a certain sign mismatch between the time translation operator and time evolution operator in quantum field theory which I hope someone can illuminate.

From my understanding, a Poincaré transformation is a transformation: $$x \to x' = \Lambda x + a,$$ for some Lorentz $\Lambda$ and vector $a$ . The transformed fields can be defined via active or passive transformation. For simplicity let us consider scalar fields only. In the passive formulation we have:

$$ \phi'(x) = \phi(\Lambda^{-1} x - \Lambda^{-1} a),$$ so that $\phi'(x') = \phi(x) $. In the active formulation we have:

$$ \phi'(x) = \phi(\Lambda x + a),$$ so that $$\phi'(x) = \phi(x').$$ Let us stick with the active formulation and postulate a unitary operator $U(\Lambda,a)$ such that the state $| \psi \rangle$ as seen in the frame $S$ corresponds to the state $| \psi' \rangle$ in the frame $S'$, i.e. we have:

$$ |\psi' \rangle = U(\Lambda,a) | \psi \rangle $$

Since we want $\langle \chi' | \phi'(x) | \psi' \rangle = \langle \chi | U^{\dagger} \phi'(x) U | \psi \rangle = \langle \chi | \phi(x) | \psi \rangle$

This gives us the identity: $$\phi'(x) = U \phi(x) U^{\dagger} = \phi(\Lambda x + a )$$

To recover Heisenberg's Equations we must postulate that $P_\mu U(\mathbb{1},a) = -i \frac{1}{\partial a^\mu} U (\mathbb{1},a)$. Thus expanding the previous equation for small $a$ and $\Lambda = \mathbb{1}$ we obtain:

$$ ( 1 + i a^\mu P_\mu) \phi(x) (1 - i a^\mu P_\mu) = \phi(x) + a^\mu \partial_\mu \phi(x)$$


$$ i [P_\mu, \phi(x) ] = \partial_\mu \phi(x).$$

In particular we have for $\mu =0$, $$i [ H,\phi(x)] = \partial_t \phi(x)$$ which is Heisenberg's equation where we identity $P_0 = H$.

Now, time translation $t \to t' = t + \Delta t$ is also a Poincaré transformation. In the active formulation which we are working now, $$\phi'(t ) = \phi( t + \Delta t ).$$ Thus in our formulation we can identify the time evolution operator $\hat{U}(t)$ via:

$$ \hat{U}(t) = U( \Lambda = \mathbb{1}, a = (t,0,0,0) ) $$

If we take this identification however, then we have by our definition of $P_\mu$, we must have:

$H \hat{U}(t) = -i \frac{d}{d t} \hat{U}(t)$, which gives the wrong sign for Schrodinger's equation, which is $$H \hat{U}(t) = +i \frac{d}{d t} \hat{U}(t).$$

Where did I go wrong?


1 Answer 1


I think I have figured it out. Everything I wrote is correct up till I identified:

$$\hat{U}(t) = U(\Lambda = 1, a^\mu = (\Delta t,0,0,0))$$

While is is true that we have:

$$U(\Lambda = 1, a^\mu = (\Delta t,0,0,0)) \phi(x) U^{\dagger}(\Lambda = 1, a^\mu = (\Delta t,0,0,0)) = \phi(x, t_0 + \Delta t )$$

Meanwhile $\hat{U}(t)$ satisfies something different. Recall from quantum mechanics that $\hat{U}(t,t_0)$, the time-evolution operator in the Shrodinger picture, is also the map between heisenberg and Shrodinger states:

$$ | \psi(t) \rangle = \hat{U}(t,t_0) | \psi(t_0) \rangle$$ (I'm taking the reference point to be $t_0$). $\ket{\psi}(t)$ is the Shrodinger state while $| \psi(t_0) \rangle$ is the heisenberg state.

The shrodinger $\phi(x,t_0)$ and heisenberg operators $\phi(x,t)$ are then related by:

$$ \phi(x,t) = \hat{U}^\dagger(t,t_0) \phi(x,t_0) U(t,t_0) $$

Now choose $\Delta t = t - t_0$. Then we have:

$$U(\Lambda = 1, a^\mu = (t-t_0,0,0,0)) \phi(x,t_0) U^{\dagger}(\Lambda = 1, a^\mu = (t-t_0,0,0,0)) = \phi(x, t )$$

Comparing the above two formulas we get:

$$ U(t,t_0) = U^{\dagger}(\Lambda = 1, a^\mu = (t-t_0,0,0,0))$$

Hence we obtain:

$$ HU(t,t_0) = i \frac{d}{dt} U(t,t_0)$$

The Shrodinger equation in the shrodinger picture. So my main confusion in my original question lies in the fact that I was working in the Heisenberg picture all along and needed to transform to the Shrodinger picture in order to reproduce Shrodinger's equation with the correct sign.


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.