The condition of equilibrium in statistical mechanics is $\frac{\partial \rho}{\partial t}=0$ where $\rho$ is the phase space density. By virtue of Liouville's theorem, this is equivalent to the statement $\{\rho,H\}=0$. Mathematically, does this mean that $H$ must also be explicitly time-independent like $\rho$ is?

  • $\begingroup$ Also be careful with explicit vs. total time dependence... $\endgroup$
    – Qmechanic
    Commented Feb 14, 2020 at 17:15
  • $\begingroup$ @ValterMoretti Thanks. So it does not mean that H have to be time-independent. $\endgroup$ Commented Feb 14, 2020 at 17:17
  • $\begingroup$ Yes, see my explicit answer. $\endgroup$ Commented Feb 14, 2020 at 17:24

1 Answer 1


If $H$ is time independent and satisfies $\{\rho, H\}=0$, also $H(t)=f(t)H+ g(t)$ does for every pairs of smooth maps $f,g$. Hence the answer is negative. $H$ can be time dependent.

  • $\begingroup$ Does the derivation of Liouville's theorem hold for a general $H$ with explicit $t$- dependence? $\endgroup$ Commented Feb 14, 2020 at 17:28
  • 2
    $\begingroup$ Yes, in the proof of the Liouville theorem it is not required that $H$ does not explicitly depend on time. $\endgroup$ Commented Feb 14, 2020 at 17:31

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.