Practical AI · Thursday, October 1, 2026
Mingyu Liu differentiates between classical simulation, based on physics equations, and neural simulation, which is data-driven. Neural simulation, used in world models, approximates predictions by learning from vast amounts of observational data, offering a pattern recognition-based approach.
“And we have simulated the environment governed by program deal on top of those equations. You know, how when two things hit each other, how things are going to happen. Those are, we consider classical simulators.”
“The scene I described earlier, people generally call it neural simulation. So it's a data-driven way of doing simulation. Instead of the diversity put the physics in, you saw tons of observation”
“It shows a lot of different scenes, visual observation to a model. And the model then can approximate, predict what's going to happen when they see similar patterns. So it's more like a pattern recognition based way of doing the simulation.”