2
$\begingroup$

I was trying to show that for an infinitesimal time evolution in classical Hamiltonian dynamics preserves volume by showing that the following Jacobian $$ \left|\frac{\partial(q(t),p(t))}{\partial(q(t_{0}),p(t_{0}))}\right| = +1 $$ By expanding $q(t)$, $p(t)$ as, $$ q(t)=q(t_{0})+\dot{q(t)}\Big|_{t_{0}} \delta t+O(\delta t^2) $$ $$ p(t)=p(t_{0})+\dot{p(t)}\Big|_{t_{0}} \delta t+O(\delta t^2) $$ using Hamilton's equation we could say $$ q(t)=q(t_{0})+\frac{\partial H}{\partial p}\Bigg|_{t_{0}} \delta t+O(\delta t^2) $$ $$p(t)=p(t_{0})-\frac{\partial H}{\partial q}\Bigg|_{t_{0}} \delta t+O(\delta t^2) $$ From the definition of jacobian matrix I get the following matrix $$\left|\begin{array}{cc} 1+\partial_{q}\partial_{p}H|_{t_{0}}\delta t+O(\delta t^2) & \partial^2 _{p}H|_{t_{0}} \delta t+O(\delta t^2)\\ -\partial^2 _{q}H|_{t_{0}} \delta t+O(\delta t^2) & 1-\partial_{p}\partial_{q}H|_{t_{0}}\delta t + O(\delta t^2) \end{array}\right|$$ Is there any other way of intuitively understanding this?

$\endgroup$
1
  • $\begingroup$ The determinant is $O(\delta t^2)$. $\endgroup$ Commented Dec 25, 2017 at 7:11

1 Answer 1

3
$\begingroup$

In view of your expansion, the temporal derivative of the Jacobian determinant, if evaluated exactly at $t=t_0$, is $$\frac{\partial J}{\partial t}|_{t_0} =\lim_{\delta t \to 0 }\frac{J(t_0+ \delta t) - J(t_0)}{\delta t}= \lim_{\delta t \to 0 }\frac{\left(1 + O(\delta t^2)\right)-1}{\delta t} =0$$ where I used Schwarz' theorem on second derivatives in computing the determinant of the Jacobian matrix you expanded, so that terms of order $\delta t$ cancel (I am assuming that the Hamiltonian function is $C^2$). Next, observe that $$J(t) = J(t_0) J(t-t_0)$$ and also $$J(t+\delta t) = J(t_0+ \delta t) J(t-t_0),$$ so that, the derivative at a generic value of time $t$ is $$\frac{\partial J}{\partial t} =\lim_{\delta t \to 0 }\frac{J(t_0+ \delta t) - J(t_0)}{\delta t} J(t-t_0) = \frac{\partial J}{\partial t}|_{t_0}J(t-t_0)= 0 J(t-t_0)=0.$$ Therefore, $J$ is constant in time for fixed canonical coordinates. As for $t=t_0$, again due to your expansion, $J(t_0)= 1$, $J(t)$ takes that value everywhere and for all times.

$\endgroup$

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.