Hello
@loudar As per usual, I enjoy reading your comments.
So I quietly changed the way the barriers work, they are much more comparable to one-way highways with easy on-ramps than they are to one-directional membranes, if that helps. The white ones are interesting, especially since they point down, and the natural tendency for hot particles is up. This can cause tight loops, so called "lit matches" as I've tended to nickname them.
As for the overall mythology of the simulation - what you mention is more-or-less what's there... almost. The black space is considered to be a compressible oil that absorbs heat (explains the auto-cooling of particles without invoking radiation). The comprehensibility of the black space goes hand-in-hand the comprehensibility of the particles themselves. They are allowed to overlap, however the more overlap there is, the greater the pressure there is - so this is where you get more energetic versus less energetic particles. In this model it couples more strongly to pressure than heat. All that said, all velocities are terminal because of the "oil". That is, all particles are normalized to move at almost the same speed. The idea that any look faster than any others is mostly an illusion. Now... all THAT said, everyone knows I hate magic numbers. Change a few variables and the whole thing changes like crazy.
Tip: decrease the time step "dt" a little bit to .5, and the whole thing acts like liquid crystal. Draw a blue line across the bottom early on.
Oh and hotkeys:
play with "b", "r", "0" and Esc
