Tiling simulation. The problem statement here is like you take a bunch of oddly shaped tiles, you don't allow them to rotate, and then you put them in a trash compactor! The trash compactor smushes the tiling down until it finds the highest density state. I'm using simulated annealing and Markov chain Monte Carlo as my trash compactor. It's not a very good trash compactor for problems like this.

The border of the applet represents the temperature (red = hot). If we set the temperature to absolute zero right away, we'd find that we don't get a very high density: we reach a "jammed" configuration and the configuration can't relax into the true highest density state. So instead, we lower the temperature ever so slowly! You should be able to see regions where the configuration "crystallizes" and it looks like we've found the highest possible density.