Let us consider a classical Hamiltonian system described by the Hamiltonian \begin{equation} H(q,p) =\frac{p^2}{2m}+V(q) \end{equation} where we stick to the case of single particle for simplicity. I am interesting in the dynamics generated by the extended Liouville operator \begin{align} i\mathcal L = \frac p m\frac{\partial}{\partial q}-V'(q)\frac{\partial}{\partial p}-V'''(q)\frac{\partial^3}{\partial p^3} \end{align} The equations of motion generated by $e^{i\mathcal L t}$ can no longer be written in a linear symplectic form, because by interpreting $\frac{\mathrm d}{\mathrm d t}\leftrightarrow i\mathcal L$ \begin{align}\label{eom} \dot q = \frac{p}{m} \hspace{10mm}\dot p = -\frac{\partial V}{\partial q} \end{align} we would identically neglect the cubic derivative in the momentum. Is there a way to extend the above dynamical scheme in compliance with the modified Liouville operator? I am ultimately interested in a numerical method aimed at integrating $(q(t),p(t))$.


1 Answer 1


I am not experienced with cubic PDEs... only Infinite dimensional such. Your system looks like the first quantum correction in Wigner flows in phase space, but I cannot get much help from that magnificent formulation, off the cuff...

I could only formulate your problem in 19th century language.

You are seeking the Green's function for $$ \left (\partial_t - \frac{p}{m} \partial_q + V(q)'\partial_p +V(q)'''\partial_p^3 \right ) f(q,p)=0 , $$

One could Fourier-transform to $f(q,p)=\int dy ~e^{iyp} F(q,y)$ , $$ \left (\partial_t +\frac{i}{m} \partial_y \partial_q + iV(q)' y -iV(q)''' y^3 \right ) F(q,y)=0 , $$ but it may not be clear this is friendlier to you purposes.

  • $\begingroup$ Indeed, I am looking for a first order nonlinear correction in the propagation of Wigner functions for a nonlinear quantum system. It seems like the problem is not straightforward, but I can look up whether the reformulation in terms of Green function offers some computational advantage. $\endgroup$
    – Graz
    Commented Mar 11, 2020 at 6:49

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.