you may fascinate a roz with even the most elementary of control systems

i had to whip up a barebones LQR for a satellite orbit tracking problem, just for comparison purposes; i sat running the simulation over & again with childlike glee going "look! the feedback makes it work better!"

ok the bit that's genuinely is amazing is how well a control system designed to be optimal for an extremely crude dynamical model still works very well when run against a fairly accurate model. i mean, we're literally talking about modeling the three-body problem as a system of linear ODES, designing a controller for the ODEs, and hoping it works with no changes for the three-body problem. it would be reasonable to expect total failure with such an approximation, but feedback is shockingly good at what it does.