1

Let's consider the unitary group $\hat{S_{\tau}^†}$ such that :$$\hat{S^†_{\tau}}|\psi(t)\rangle=|\psi(t-\tau)\rangle$$ Since we know that: $$\hat{U}(t,t_0)|\psi(t_0)\rangle=|\psi(t)\rangle$$ Where ${U}(t,t_0)$ is the time evolution operator, we can conclude that: $$\hat{S^†_{\tau}}=U(t-\tau,t).$$ Now we also know that $\hat{U}$ solves: $$iℏ\partial_t\hat{U}=\hat{H}\hat{U}$$ So that if $\hat{H}$ does not depend explicitly on time we have: $$U(t,t_0)=e^{-\frac{i}{ℏ}(t-t_0)\hat{H}} (*)$$ Hence: $$\hat{S^†_{\tau}}=U(t-\tau,t)=e^{\frac{i}{ℏ}\tau\hat{H}}$$ $$\hat{S_{\tau}}=e^{-\frac{i}{ℏ}\tau\hat{H}}$$ And we can surely say that $\hat{H}$ is the generator of time translations. Now what happens when $\hat{H}$ has an explicit dependence on time ? Equation $(*)$ no longer holds and I am not confident in saying that $\hat{H}$ is still the generator of the group $\hat{S_{\tau}}$. The book I'm reading doesn't distinguish between these two cases. Is it because it only considers infinitesimal translations?

Qmechanic
  • 201,751
davise
  • 41
  • 2
    This depends on how you define "generator". In the case where $H$ depends on time, Stone's theorem does not hold anymore. – Tobias Fünke Mar 18 '24 at 19:26
  • Is it possible to define a generator without referring to Stone's theorem ? – davise Mar 18 '24 at 19:30
  • What do you mean with "generator"? – Tobias Fünke Mar 18 '24 at 19:39
  • Review time-ordered exponentials. It should become evident to you what "generator", an infelicitous term here, represents. – Cosmas Zachos Mar 18 '24 at 20:12
  • Linked and also. "Generator" is shorthand for "Agent", as in sufficient determinator. You may think of an infinity of generators, one for each time instant. – Cosmas Zachos Mar 18 '24 at 20:17
  • The instantaneous generator of infinitesimal time translations at time $t$ is $H(t)$. There is no single generator for all time because a time-dependent evolution doesn't enjoy time-translation invariance. Accordingly, time evolution can't be generated by a single operator in the sense of Lie algebras and groups (i.e. your $\hat{S}$ guy is no longer a [semi]group). Most people wouldn't use the term "generator" in this context because of that. – just a phase Mar 29 '24 at 03:29

1 Answers1

0

Consider the solution of: $$i\hbar{\partial\over \partial t}\mathcal U(t,t_0)=H\mathcal U(t,t_0).$$ If the Hamiltonian is time dependent and if it commutes with itself at different times, $$[H(t),H(t^\prime)]=0,\;t\neq t^\prime.$$ Then the time development is given by: $$\mathcal U(t,t_0)=e^{-{i\over\hbar}\int_{t_0}^t H(t^\prime)\;dt^\prime}.$$ If the Hamiltonian does not commute with itself at different times: $$[H(t),H(t^\prime)]\neq 0,$$ then the time development operator is given as: $$\mathcal U(t,t_0)=1+\sum\limits_{n=1}\limits^{\infty}\bigg({-i\over\hbar}\bigg)^n\int_{t_0}^t\;dt_1\int_{t_0}^{t_1}\;dt_2\cdots \int_{t_0}^{t_{n-1}}\;dt_n\;H(t_1)H(t_2)\cdots H(t_n). $$ This last result is known as a Dyson series and is important in quantum field theory.