I'm currently consuming a course on QFT where we need to define the unitary time-evolution to get the time evolution of the wave function in the interaction picture:

$\hat{U}(t_1,t_0) = \exp\left(\frac{i}{\hbar}\hat{H}_0t_1\right)\exp\left(-\frac{i}{\hbar}\hat{H}(t_1-t_0)\right)\exp\left(-\frac{i}{\hbar}\hat{H}_0t_0\right)$ .

Now one can show that this operator follows a Schrödinger equation by simply taking the derivative to time:

$i\hbar\frac{d}{dt}\hat{U}(t,t_0) = \hat{H}_1^I(t)\hat{U}(t,t_0)$ .

Where $\hat{H}_1^I$ is the perturbation to our free-field Hamiltonian $\hat{H}_0$.

Now I started wondering whether $\hat{U}(t_1,t_0)$ shouldn't also follow a Heisenberg equation since it's an operator.

I believe it shouldn't since $\hat{U}(t_1,t_0)$ gives a unitairy time-evolution which is a transformation, while the Heisenberg equation applies to observables. I was woundering if someone could confirm my reasoning or disprove it?

  • $\begingroup$ If $t_1$ and $t_0$ are constants, then $\hat{U}(t_1,t_0)$ is the operator evaluated at those particular instants, not a function. Think about the function $\sin(x)$ and the function evaluated at $x_0$ $\sin(x)|_{x=x_0}=\sin(x_0)$ $\endgroup$
    – J L
    Jun 24, 2013 at 14:30
  • $\begingroup$ Yes but if I were to take $t_1$ as my variable, then it might follow some equation governing the time-evolution. That it follows Schrödinger can be proven. For the Heisenberg-case this doesn't work (I wouldn't know how to take the commutator of $\hat{H_1^I}$ and $\hat{H_0}$). My reasoning for this is that the Heisenberg-equation only applies to observables, not the transformation (what the $\hat{U}(t_1,t_0)$ in principle is. $\endgroup$
    – Nick
    Jun 24, 2013 at 14:54

1 Answer 1


Yes, OP is right. The Heisenberg evolution equation in the interaction picture applies to an operator $A_I(t)$, that depends on a single moment of time $t$. On the other hand, the evolution operator $\hat{U}(t_f,t_i)$ depends in principle on the whole intermediate time interval $[t_i,t_f]$, and is in particular not a local operator of a single time.

  • $\begingroup$ Ok, that might explain my conceptual problems upon trying to derive the Heisenberg-like equation. Thanks! Now I was woundering if it was purely because of the fact that $\hat{U}(t_f,t_i)$ is depending on the interval $[t_i,t_f]$. Or also because of the fact that it is a transformation and Heisenberg doesn't apply to transformations ? $\endgroup$
    – Nick
    Jun 24, 2013 at 15:20
  • 1
    $\begingroup$ @Nick: Well, I'm not sure what class of transformations you allow, but a sufficiently well-behaved local transformation operator that is only associated with a single time should satisfy the Heisenberg evolution equation. $\endgroup$
    – Qmechanic
    Jun 24, 2013 at 15:41
  • $\begingroup$ @Nick: An other way to see that is that the evolution operator have to be expressed as a time ordered function of the perturbation hamiltonian. $\endgroup$
    – Trimok
    Jun 24, 2013 at 17:00
  • $\begingroup$ Why would they violate Heisenberg if I'd have a time-ordered product ? $\endgroup$
    – Nick
    Jun 24, 2013 at 18:05
  • 1
    $\begingroup$ @Nick : Yes. Be careful, it is a partial derivative. (Minor remark : if you want your message appears in the mail box of somebody, you have to do a @ (for instance @Trimok). Sometimes it is automatic, but not in this case.) $\endgroup$
    – Trimok
    Jun 25, 2013 at 18:21

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.