Open GiudGiud opened 3 years ago
I've had this happen in very rare cases as well. You can set the parameter timestep_tolerance
in the Executioner
to a value a little higher than the default value (2e-14) and that has helped me in these situations. Especially when you are taking 1 second timesteps I would increase the default a few orders of magnitude and see if that works.
I believe there is a timestep_tolerance
parameter that gets around this, but i completely agree that this sort of thing is stupid
Bug Description
Constant time stepper struggles on last time step. Very likely to be a numerical round off issue. Cant really disallow this case. Maybe we should have a default saying "hey if it s within 1% of what we would have taken we ll just do 101% and call it a day"
Steps to Reproduce
Run the VTB msr/steady inputs
Impact
Annoying