It's pretty cool to see instability in this context, it seems similar to the idea of compounding errors in reinforcement learning where your trajectory diverges instead of the simulation.
JaxonZeng
The discrete factor of our computers means that we cannot truly simulate the real physical world. The errors will build up after time steps and make our simulation unrealistic. This can thus cause instability since the computation results can round up and cause the energy to increase.
antony-zhao
I wonder if this would technically fall under the idea of "chaos theory" where very tiny differences would cause 2 things to diverge significantly, where in this case it'd be the very small numerical differences cause the simulation to deviate from reality.
It's pretty cool to see instability in this context, it seems similar to the idea of compounding errors in reinforcement learning where your trajectory diverges instead of the simulation.
The discrete factor of our computers means that we cannot truly simulate the real physical world. The errors will build up after time steps and make our simulation unrealistic. This can thus cause instability since the computation results can round up and cause the energy to increase.
I wonder if this would technically fall under the idea of "chaos theory" where very tiny differences would cause 2 things to diverge significantly, where in this case it'd be the very small numerical differences cause the simulation to deviate from reality.