No this is actually using neural nets to predict and control plasma boundary disruptions, which is an intractible problem to solve numerically.
I've been studying plasma physics and from what I understand perfect control isn't necessary as long as you can control it long enough to get useful power. If the plasma dissipates you just restart. But ideally it's controlled enough that while it is in there, it's producing net energy
> an intractible problem to solve numerically.
Yeah, so we'll just get a reduced dimension model that we don't know how it works.
Sounds good.
Is "producing net energy" really what’s physically happening here, or is it a shortcut for "extracting pre-existing matter/energy in a form we can canalize?