Equilibrium Stat Mech Vs. Kinetics

dimanche 13 juillet 2014

Hey guys, I'm extremely confused about the distinction between equilibrium statistical mechanics and kinetics, based on literally just the first section of vol. 5 and vol. 10 of Landau.



Starting off with (classical) phase space (for simplicity), for a system of [itex]n[/itex] particles with [itex]3n[/itex] coordinate degrees of freedom and [itex]3n[/itex] momenta we have a [itex]6n[/itex] dimensional phase space [itex]\Gamma[/itex] containing points [itex](\vec{q},\vec{p})[/itex] (known as micro-states). The motion of the system over a time interval is represented as a parametrized curve [itex]\gamma[/itex], [itex]\gamma (t) = (\vec{q}(t),\vec{p}(t)) = (\vec{q},\vec{p}) [/itex]. The form of the curve [itex]\gamma[/itex] for a physical system is determined by the Hamiltonian, but what is actually going on here? Do I say that [itex]H = H(\vec{q}(t),\vec{p}(t)) = E(t)[/itex] determines the motion, or do I say that [itex]H = H(\vec{q}(t),\vec{p}(t),t) = E(t)[/itex] determines the motion? The two situations are different.



Is it that [itex]H = H(\vec{q}(t),\vec{p}(t)) = E(t)[/itex] determines the motion of the system (the curve [itex]\gamma[/itex]), which in general has no dimensional reduction or restrictions (i.e. we have [itex]6n[/itex] degrees of freedom), but that, in the time-independent case [itex]H = H(\vec{q}(t),\vec{p}(t)) = E[/itex] we have only [itex]6n-1[/itex] free coordinates (should I maybe write [itex]H = H(\vec{q},\vec{p}) = E[/itex]?), resulting in our curve living on the energy shell in the picture:







Or, is it that [itex]H = H(\vec{q}(t),\vec{p}(t),t) = E(t)[/itex] determines the motion in general, and we can draw the above picture in the general case, however the [itex]6n-1[/itex] dimensional surface changes with time? In other words, adding an extra [itex]t[/itex] coordinate does not change the dimensionality of phase space, rather it parametrizes the energy surface, allowing it to vary in time. Only in the time-independent case [itex]H = H(\vec{q}(t),\vec{p}(t)) = E[/itex] do we have a constant-energy surface, but even in the time-dependent case we can draw a [itex]6n-1[/itex] dimensional surface on which the curve lives, so long as the surface changes with time.



So far, we just have a curve in [itex]\Gamma[/itex] space, no statistical mechanics or kinetics.




Attached Images





File Type: png Energy.png (4.0 KB)







0 commentaires:

Enregistrer un commentaire