[FLASH-USERS] parameters changes allowed on restart?

[dongwook at flash.uchicago.edu]

Dear Andrew,

The original implementation works correctly for a normal situation
in which dr_dt is also a value that is being read from a checkpoint
file for a restart.

Therefore, it is an expected behavior that you saw your run got restarted
with the old small timesteps, because that was the value you used
in your previous run, and got checkpointed.

Note that dtmax is a ceiling value (upper limit) that keeps your dr_dt
from growing more than what you expect, and I don't think this is
not what you wanted: I think what you want to have is "dtmin"
(a floor value) in order to restart with a larger dt than dr_dt.

Best,
Dongwook

> Okay, I managed to get the integration step size to begin auto changing
> again by replacing the code in
>
> source/physics/Hydro/HydroMain/split/MHD_8WV/Hydro_init.F90
>
> that says:
>
>   if (.not.dr_restart) then
>      hy_dtmin = HUGE(hy_dtmin)
>   else
>      hy_dtmin = dr_dt
>   endif
>
> with just:
>
> hy_dtmin = HUGE(hy_dtmin)
>
> That is, with my combination of units, I think the driver is not
> properly setting dr_dt on restarts - I'll leave it to the experts to
> decide how this should really be fixed...
>
> Thanks,
>      Andy S.
>
> John ZuHone wrote:
>> Andrew,
>>
>> So long as the timestep is not restricted by either dtmin or dtmax it
>> will always be whatever the CFL condition sets it to, so long as the
>> only physics included in the simulation that sets timesteps is hydro. So
>> if it is less than your current dtmax increasing dtmax will not change
>> the behavior.
>>
>>
>