Interesting approaches to crowd simulation by abstracting agents to fluid fields (around 6:20), and model reduction for fast simulation of high-dimensional fluid problems (around 23:00) and realtime control (33:00):
I haven’t really digested the implications of this, but it’s interesting to consider what the implications might be for simulating lumpier systems, like traditional SD or economic models, where model reduction has not been very widespread, or for large-scale computing like climate models.