Hover over the menu bar to pick a physics animation.


Maxwell's Demon Energy Seperator

The idea here is to use intelligent knowledge of energies in order to segregate the higher energy particles to the right side of the chamber. All the particles start on the left side but, if the energy of one of these is larger than the threshold, Et, then it is allowed to transit to the right side and be trapped there (and have its color changed from black to blue). If you turn on the sound of your computer, you can hear a click when the discs cross from left to right side of the chamber. This, of course, violates the second law of thermodynamics and can only be done by a priori knowing the energy of particles impinging on the divider between the left and right side. Most instances of the second law are due to molecular collisions or other molecular interactions. A large fraction of the controversy about the second law is that Newton's collision laws are perfectly valid when you view the collisions time reversed. That means that the second law can have exceptions so that reduces it to a statistical law meaning that the time-reversed case is just very unlikely but I think almost everyone would go along with that anyway. For more information about this see Photonic Maxwell Demon For the left side of the chamber, the number of discs available to move to the right side is

`int_(E_t)^oo[exp(-E/E_(avg))dE]/E_(avg)`

with limits from `E_t` to infinity. Using this integral, the rate of transfer of these discs is

`(dn)/dt=E_t/E_("Leftavg")(n_("Left")v_("Left"))/(2L)exp(-E_t/E_("Left"))`

where `L` is the width of the chamber's left side, `v_("Left")` is the average speed of the discs on the left side, `E` denotes energy, and `n` denotes the number of discs. The red curves shows the actual number transferred and the green curve shows the value estimated from this equation. The actual energy distributions are plotted in black (left) and blue (right) as histograms. These histograms eventually fit the corresponding expected exponential curves (exp(-E/) for a 2D energy distribution. To see that the fit is good requires that the program run at least several minutes.